In [1]:
from IPython.core.display import display, HTML
display(HTML("<style>.container { width:100% !important; }</style>"))

import os
import pandas as pd
import missingno as msno
import seaborn as sns
from pandas.plotting import scatter_matrix
import matplotlib.pyplot as plt
import numpy as np
from collections import defaultdict
from sklearn.model_selection import GroupShuffleSplit, ShuffleSplit, LeaveOneOut, GridSearchCV
from sklearn import linear_model
from sklearn.tree import _tree, export_graphviz, DecisionTreeRegressor
from sklearn.metrics import mean_squared_error, median_absolute_error
from sklearn.externals import joblib
import graphviz
import numpy as np
import seaborn as sns
%matplotlib inline

pd.options.display.max_columns=100
pd.options.display.max_rows=100


Creation Of Model


In [2]:
df = pd.read_csv('cleaned_df4.csv')
df = df.drop('gallons per bedroom per month', axis=1)
cols = df.columns.tolist()
cols = [col for col in cols if not col.endswith('.1')]
df = df[cols]
df = df.rename(columns={'match building name': 'building'})
d = df.copy()

#### Beginning of outlier detection

df = df[['building', 'month', 'year', 'target']]
month_mapping = {'april': 4,
 'august': 8,
 'december': 12,
 'february': 2,
 'january': 1,
 'july': 7,
 'june': 6,
 'march': 3,
 'may': 5,
 'november': 11,
 'october': 10,
 'september': 9}

df['month_num'] = df['month'].map(month_mapping)

df['date'] = ['{}/1/{}'.format(month, year) for month, year in df[['month_num', 'year']].values]
df['date'] = pd.to_datetime(df['date'], format='%m/%d/%Y')

bldg_dict = {}

yes_count, no_count = 0, 0

stddev_n = 3

months = []

total_outliers = []

master = pd.DataFrame()

for building, frame in df.groupby('building'):
    frame = frame.set_index('date', inplace=False).sort_index()
    med = np.median(frame['target'])
    std = np.std(frame['target'])

    lower, upper = -stddev_n*std, stddev_n*std

    lower_cons, upper_cons = (med + lower, med + upper)

    lower_cons = lower_cons if lower_cons >= 0 else 0

    bldg_dict[building] = {'med': med, 'std': std, 'lower_cons': lower_cons, 'upper_cons': upper_cons}
    cons = frame['target'].values.tolist()

   
    if any(obs > upper_cons or obs < lower_cons for obs in cons):
        yes_count += 1
        filt = frame.loc[~(frame['target'] > upper_cons) | (frame['target'] < lower_cons)]

        filt['outliers_removed'] = True
        master = pd.concat([master, filt])
        # print('Outliers detected for {}'.format(building))

    else:
        no_count += 1
        frame = frame.assign(outliers_removed = False)
        # print('No outliers detected for {}'.format(building))
        master = pd.concat([master, frame]) 
print('Number of buildings with outliers:  {}'.format(yes_count))
print('Number of buildings without outliers:  {}'.format(no_count))


/Users/Greg/anaconda/envs/main/lib/python3.4/site-packages/ipykernel/__main__.py:61: SettingWithCopyWarning: 
A value is trying to be set on a copy of a slice from a DataFrame.
Try using .loc[row_indexer,col_indexer] = value instead

See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy
Number of buildings with outliers:  47
Number of buildings without outliers:  30

In [3]:
for building, group in master.groupby('building'):
    plt.scatter(x=group.index, y=group['target'])



In [4]:
d_ = pd.merge(d, master, left_on=['building', 'month', 'year', 'target'], right_on=['building', 'month', 'year', 'target'], how='right')

In [5]:
d_.shape


Out[5]:
(4205, 42)

Create Feature Matrix


In [11]:
df = d_.copy()

In [12]:
cols = df.columns.tolist()
cols = [col for col in cols if df[col].dtypes == 'object']
for col in cols:  df[col] = df[col].str.lower()

In [15]:
df.head()


Out[15]:
building month year single family? single family building type low-income housing predominant resident type type of construction is there a basement? is basement finished & heated? environmental certification if environmental certification: type heating fuel heating system hot water system fuel hot water system cooling system common laundry facilities if common laundry, dryer fuel total # bedrooms in building target gross bldg sq ft if multi-family sum of apt sq ft basement sq ft # stories if multi-family, # units # elevators # of days in month 0 bedrooms - days vacant 1 bedrooms days vacant 2 bedrooms days vacant 3 bedrooms days vacant 4 bedrooms days vacant # of 0 bedrooms * days in month # of 1 bedrooms * days in month # of 2 bedrooms * days in month # of 3 bedrooms * days in month # of 4 bedrooms * days in month # of 5 bedrooms * days in month bldg_age month_num outliers_removed
0 1822 park august 2017 no 0 yes mixed/none masonry yes yes no no natural gas boiler (steam) natural gas other window ac yes natural gas 18 30862.9998 18000.0 9772 4500 4.0 18 0 31 0 0 0 0 0 0 558 0 0 0 0 101 8 True
1 alliance addition-730 17th street august 2017 no 0 yes mixed/none wood or steel frame yes yes yes leed, platinum electricity heat pump (air-source) electricity stand-alone storage water heater ducted central ac (outdoor condenser) yes natural gas 61 78751.0000 33124.0 20790 7126 3.0 51 1 31 261 0 0 0 0 1829 62 0 0 0 0 117 8 False
2 archdale august 2017 no 0 yes mixed/none masonry yes yes no no natural gas boiler (steam) natural gas stand-alone storage water heater window ac yes natural gas 30 117300.0000 19760.0 12178 4940 3.0 30 0 31 48 33 0 0 0 806 124 0 0 0 0 98 8 True
3 st. barnabas august 2017 no 0 yes mixed/none masonry yes yes no no natural gas boiler (hot water) natural gas indirect hot water tank off boiler (heat & dhw) ducted central ac (outdoor condenser) yes natural gas 52 88704.6004 32467.0 17316 6839 5.0 52 1 31 277 0 0 0 0 1612 0 0 0 0 0 12 8 True
4 buri manor august 2017 no 0 yes mixed/none wood or steel frame no no no no electricity electric baseboard electricity stand-alone storage water heater window ac yes natural gas 38 63680.7990 16464.0 9970 0 3.0 38 0 31 48 0 0 0 0 1178 0 0 0 0 0 31 8 True

In [16]:
# Need to split up the dataframes into categoricals and numericals
cats = df.loc[:, 'month':'if common laundry, dryer fuel']
cats = cats.drop('year', axis=1)
nums = df.loc[:, 'total # bedrooms  in building':'bldg_age']
nums = nums.drop('target', axis=1)
targets = df['target'].copy()
cats.head()


Out[16]:
month single family? single family building type low-income housing predominant resident type type of construction is there a basement? is basement finished & heated? environmental certification if environmental certification: type heating fuel heating system hot water system fuel hot water system cooling system common laundry facilities if common laundry, dryer fuel
0 august no 0 yes mixed/none masonry yes yes no no natural gas boiler (steam) natural gas other window ac yes natural gas
1 august no 0 yes mixed/none wood or steel frame yes yes yes leed, platinum electricity heat pump (air-source) electricity stand-alone storage water heater ducted central ac (outdoor condenser) yes natural gas
2 august no 0 yes mixed/none masonry yes yes no no natural gas boiler (steam) natural gas stand-alone storage water heater window ac yes natural gas
3 august no 0 yes mixed/none masonry yes yes no no natural gas boiler (hot water) natural gas indirect hot water tank off boiler (heat & dhw) ducted central ac (outdoor condenser) yes natural gas
4 august no 0 yes mixed/none wood or steel frame no no no no electricity electric baseboard electricity stand-alone storage water heater window ac yes natural gas

In [20]:
nums.columns.tolist()  # Numerical Variables Kept For Modeling
cats.columns.tolist()  # Categorical variables to keep for modeling
cats_ = pd.get_dummies(cats)  # Encode the categorical variables

# Concatenate the targets, the buildings, the categorical variables, and the numerical variables, then create a holdout training / validation split
bldgs = df['building']
X_df = pd.concat([ bldgs, targets, cats_, nums], axis=1)
valid_set = X_df.sample(frac=0.2)
X_df = X_df.loc[~X_df.index.isin(valid_set.index)].copy()

print('Size of the dataset:  {}'.format(X_df.shape))
print('Size of the validation set:  {}'.format(valid_set.shape))
print('Size of the new training set is:  {}'.format(X_df.shape))
print('Describe the training dataset...')
print(X_df['target'].describe())


Size of the dataset:  (3364, 85)
Size of the validation set:  (841, 85)
Dropping the validation samples...
Size of the new training set is:  (3364, 85)
Describe the training dataset...
count      3364.000000
mean      98305.166893
std      102565.203001
min           0.000000
25%       21066.000000
50%       69099.949400
75%      138104.498625
max      537901.005000
Name: target, dtype: float64

In [21]:
X_df.head()


Out[21]:
building target month_april month_august month_december month_february month_january month_july month_june month_march month_may month_november month_october month_september single family?_no single family?_yes single family building type_0 single family building type_detached single family building type_shared wall low-income housing_no low-income housing_yes predominant resident type_elderly predominant resident type_mixed/none type of construction_masonry type of construction_wood or steel frame is there a basement?_no is there a basement?_yes is basement finished & heated?_no is basement finished & heated?_yes environmental certification_no environmental certification_yes if environmental certification: type_leed, gold if environmental certification: type_leed, platinum if environmental certification: type_no if environmental certification: type_other (type in) heating fuel_electricity heating fuel_natural gas heating fuel_steam heating system_boiler (high efficiency condensing) heating system_boiler (hot water) heating system_boiler (steam) heating system_electric baseboard heating system_furnace heating system_furnace (high efficiency condensing) heating system_heat pump (air-source) heating system_heat pump (ground-source) heating system_ptac units hot water system fuel_electricity hot water system fuel_natural gas hot water system_indirect hot water tank off boiler (heat & dhw) hot water system_indirect hot water tank off dedicated boiler hot water system_no hot water system_other hot water system_stand-alone storage water heater cooling system_air-cooled chiller cooling system_ducted central ac (outdoor condenser) cooling system_packaged rooftop cooling system cooling system_ptac unit cooling system_room ac in sleeve (through wall ) cooling system_window ac common laundry facilities_no common laundry facilities_yes if common laundry, dryer fuel_0 if common laundry, dryer fuel_electricity if common laundry, dryer fuel_natural gas total # bedrooms in building gross bldg sq ft if multi-family sum of apt sq ft basement sq ft # stories if multi-family, # units # elevators # of days in month 0 bedrooms - days vacant 1 bedrooms days vacant 2 bedrooms days vacant 3 bedrooms days vacant 4 bedrooms days vacant # of 0 bedrooms * days in month # of 1 bedrooms * days in month # of 2 bedrooms * days in month # of 3 bedrooms * days in month # of 4 bedrooms * days in month # of 5 bedrooms * days in month bldg_age
0 1822 park 30862.9998 0 1 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 1 0 1 1 0 0 1 0 1 1 0 0 0 1 0 0 1 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 1 0 1 0 0 1 18 18000.0 9772 4500 4.0 18 0 31 0 0 0 0 0 0 558 0 0 0 0 101
2 archdale 117300.0000 0 1 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 1 0 1 1 0 0 1 0 1 1 0 0 0 1 0 0 1 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 1 0 1 0 0 1 30 19760.0 12178 4940 3.0 30 0 31 48 33 0 0 0 806 124 0 0 0 0 98
3 st. barnabas 88704.6004 0 1 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 1 0 1 1 0 0 1 0 1 1 0 0 0 1 0 0 1 0 0 1 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 1 52 32467.0 17316 6839 5.0 52 1 31 277 0 0 0 0 1612 0 0 0 0 0 12
5 chicago ave-1500 113390.0012 0 1 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 1 0 1 1 0 0 1 0 1 1 0 0 0 1 0 0 1 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 1 0 1 0 0 1 28 21328.0 15993 7109 3.0 24 0 31 0 0 0 0 0 0 682 0 62 0 0 101
7 chicago ave-1508 77386.8004 0 1 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 1 0 1 1 0 0 1 0 1 1 0 0 0 1 0 0 1 0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 1 0 1 0 0 1 26 17532.0 13104 5844 3.0 12 0 31 0 0 0 0 0 0 0 310 62 0 0 101

In [22]:
sns.distplot(X_df['target'])


/Users/Greg/anaconda/envs/main/lib/python3.4/site-packages/statsmodels/nonparametric/kdetools.py:20: VisibleDeprecationWarning: using a non-integer number instead of an integer will result in an error in the future
  y = X[:m/2+1] + np.r_[0,X[m/2+1:],0]*1j
Out[22]:
<matplotlib.axes._subplots.AxesSubplot at 0x116594a90>

In [23]:
def create_dsets(df):
    target = df['target'].values
    bldgs = df['building'].tolist()
    X = df.drop(['target', 'building'], axis=1)
    features = X.columns.tolist()
    X = X.values
    return X, target, bldgs, features

In [38]:
def makeplot(estimator, X, y, X_valid, y_valid, title):
    predicted = estimator.predict(X)
    pred_valid = estimator.predict(X_valid)
    
    score = "Training R2 Score:  {}".format(estimator.score(X, y))
    score2 = "Validation R2 Score:  {}".format(estimator.score(X_valid, y_valid))

    mse = "Training MSE:  {}".format(mean_squared_error(y, estimator.predict(X)))
    mse2 = "Validation MSE:  {}".format(mean_squared_error(y_valid, estimator.predict(X_valid)))

    mae = 'Training Median Absolute Error is:  {}'.format(median_absolute_error(y, estimator.predict(X)))
    mae2 = 'Validation Median Absolute Error is:  {}'.format(median_absolute_error(y_valid, estimator.predict(X_valid)))

    plt.figure(figsize=(15,9))
    plt.title(title)
    
    plt.scatter(y, y)
    plt.scatter(y, predicted, edgecolors='red', facecolors='none')
    plt.scatter(y_valid, pred_valid, edgecolors='green', facecolors='green')
    
    plt.annotate(score, xy=(0.05, .86), xycoords='axes fraction')
    plt.annotate(score2, xy=(.05, .84), xycoords='axes fraction')
    
    plt.annotate(mse, xy=(.05, .82), xycoords='axes fraction')
    plt.annotate(mse2, xy=(.05, .80), xycoords='axes fraction')


    plt.annotate(mae, xy=(.05, .78), xycoords='axes fraction')
    plt.annotate(mae2, xy=(.05, .76), xycoords='axes fraction')


    plt.annotate('Blue = Actual', xy=(.05, .90), xycoords='axes fraction')
    plt.annotate('Red = Predicted', xy=(.05, .88), xycoords='axes fraction')
    plt.annotate('Green = Validation', xy=(0.05, 0.92), xycoords='axes fraction')
    
    plt.xlabel('Actual Water Consumption per Building')
    plt.ylabel('Predicted Water Consumption per Building')
    plt.show()

In [39]:
def train_model(X, y, estimator, params):
    grid = GridSearchCV(estimator = estimator, cv=10, param_grid=params, n_jobs=-1)
    grid.fit(X, y)
    return grid

In [73]:
def plot_variables(features, data, X):
    """Describes the number of features used in model training and their importance
    
    :param features = A list of feature names
    :param data = The coefficients of feature_importances associated with the best model
    :param X = The array used for model training
    """
    
    print('The model was trained with {} features.  Of those {} features contribute to the outcome.'.format(X.shape[1], len(data)))
    f = pd.DataFrame(index=features, data=data, columns=['Feature Importance'])
    f.loc[f_['Strength'] != 0].plot(kind='barh', figsize=(10,10))
    return None

In [74]:
# Create the matrices for training a model, the target values, list of building names, and the featurenames for both the training data, and the validation data
X, y, bldgs, features = create_dsets(X_df)
X_valid, y_valid, bldgs_valid, _ = create_dsets(valid_set)

In [75]:
features


Out[75]:
['month_april',
 'month_august',
 'month_december',
 'month_february',
 'month_january',
 'month_july',
 'month_june',
 'month_march',
 'month_may',
 'month_november',
 'month_october',
 'month_september',
 'single family?_no',
 'single family?_yes',
 'single family building type_0',
 'single family building type_detached ',
 'single family building type_shared wall ',
 'low-income housing_no',
 'low-income housing_yes',
 'predominant resident type_elderly',
 'predominant resident type_mixed/none',
 'type of construction_masonry',
 'type of construction_wood or steel frame',
 'is there a basement?_no',
 'is there a basement?_yes',
 'is basement finished & heated?_no',
 'is basement finished & heated?_yes',
 'environmental certification_no',
 'environmental certification_yes',
 'if environmental certification: type_leed, gold',
 'if environmental certification: type_leed, platinum',
 'if environmental certification: type_no',
 'if environmental certification: type_other (type in)',
 'heating fuel_electricity',
 'heating fuel_natural gas',
 'heating fuel_steam',
 'heating system_boiler (high efficiency condensing) ',
 'heating system_boiler (hot water)',
 'heating system_boiler (steam)',
 'heating system_electric baseboard',
 'heating system_furnace',
 'heating system_furnace (high efficiency condensing) ',
 'heating system_heat pump (air-source)',
 'heating system_heat pump (ground-source)',
 'heating system_ptac units',
 'hot water system fuel_electricity',
 'hot water system fuel_natural gas',
 'hot water system_indirect hot water tank off boiler (heat & dhw)',
 'hot water system_indirect hot water tank off dedicated boiler ',
 'hot water system_no',
 'hot water system_other',
 'hot water system_stand-alone storage water heater',
 'cooling system_air-cooled chiller',
 'cooling system_ducted central ac (outdoor condenser) ',
 'cooling system_packaged rooftop cooling system',
 'cooling system_ptac unit',
 'cooling system_room ac in sleeve (through wall )',
 'cooling system_window ac ',
 'common laundry facilities_no',
 'common laundry facilities_yes',
 'if common laundry, dryer fuel_0',
 'if common laundry, dryer fuel_electricity',
 'if common laundry, dryer fuel_natural gas',
 'total # bedrooms  in building',
 'gross bldg sq ft',
 'if multi-family sum of apt sq ft',
 'basement sq ft',
 '# stories',
 'if multi-family, # units',
 '# elevators',
 '# of days in month',
 '0 bedrooms - days vacant',
 '1 bedrooms days vacant',
 '2 bedrooms days vacant',
 '3 bedrooms days vacant',
 '4 bedrooms days vacant',
 '# of 0 bedrooms * days in month',
 '# of 1 bedrooms * days in month',
 '# of 2 bedrooms * days in month',
 '# of 3 bedrooms * days in month',
 '# of 4 bedrooms * days in month',
 '# of 5 bedrooms * days in month',
 'bldg_age']

Benchmark Linear Regression


In [76]:
lr_ = linear_model.LinearRegression()
lr = train_model(X=X, y=y, estimator=lr_, params=[{'normalize': [True, False]}])
makeplot(estimator=lr, X=X, y=y, X_valid=X_valid, y_valid=y_valid, title='Linear Regression 10-Fold Cross-Validation')
plot_variables(features=features, data=lr.best_estimator_.coef_, X=X)


The model was trained with 83 features.  Of those 83 features contribute to the outcome.

Decision Tree Regressor


In [85]:
clf_ = DecisionTreeRegressor(criterion='mae')
clf1 = train_model(X=X, y=y, estimator=clf_, params=[{'max_depth': [6], 'min_samples_split': [4, 6]}])
makeplot(estimator=clf1, X=X, y=y, X_valid=X_valid, y_valid=y_valid, title='Decision Tree Regression with 10-Fold Cross-Validation')
plot_variables(features=features, data=clf1.best_estimator_.feature_importances_, X=X)


The model was trained with 83 features.  Of those 83 features contribute to the outcome.

In [86]:
clfplot=export_graphviz(clf1.best_estimator_, out_file='my_tree.dot', feature_names=features)
with open('my_tree.dot') as f:
    dot_graph=f.read()
graphviz.Source(dot_graph)


Out[86]:
Tree 0 total # bedrooms  in building <= 53.0 mae = 72424.854 samples = 3364 value = 69099.949 1 total # bedrooms  in building <= 24.0 mae = 28218.987 samples = 2261 value = 31225.2 0->1 True 64 total # bedrooms  in building <= 126.5 mae = 77549.295 samples = 1103 value = 185532.998 0->64 False 2 total # bedrooms  in building <= 8.5 mae = 12237.745 samples = 1509 value = 18835.5 1->2 33 if multi-family, # units <= 25.5 mae = 19269.76 samples = 752 value = 83264.15 1->33 3 single family?_yes <= 0.5 mae = 5861.254 samples = 903 value = 13479.5 2->3 18 single family building type_shared wall  <= 0.5 mae = 11378.372 samples = 606 value = 34499.352 2->18 4 # stories <= 2.25 mae = 4785.348 samples = 680 value = 15622.8 3->4 11 total # bedrooms  in building <= 5.0 mae = 3370.972 samples = 223 value = 4163.98 3->11 5 gross bldg sq ft <= 3079.5 mae = 4579.459 samples = 632 value = 16099.9 4->5 8 4 bedrooms days vacant <= 1.0 mae = 859.804 samples = 48 value = 8412.37 4->8 6 mae = 5290.468 samples = 130 value = 11430.95 5->6 7 mae = 3908.014 samples = 502 value = 16903.35 5->7 9 mae = 727.24 samples = 43 value = 8571.62 8->9 10 mae = 412.592 samples = 5 value = 6604.85 8->10 12 # of 3 bedrooms * days in month <= 14.0 mae = 2321.16 samples = 176 value = 3189.86 11->12 15 bldg_age <= 23.5 mae = 1627.534 samples = 47 value = 10400.2 11->15 13 mae = 1070.696 samples = 93 value = 2075.1 12->13 14 mae = 2623.43 samples = 83 value = 5723.81 12->14 16 mae = 1859.206 samples = 19 value = 13040.3 15->16 17 mae = 941.73 samples = 28 value = 9869.77 15->17 19 cooling system_window ac  <= 0.5 mae = 11227.654 samples = 480 value = 36983.65 18->19 26 # of 0 bedrooms * days in month <= 29.5 mae = 4763.505 samples = 126 value = 24099.65 18->26 20 single family building type_0 <= 0.5 mae = 12287.923 samples = 177 value = 48184.8 19->20 23 single family building type_detached  <= 0.5 mae = 7579.429 samples = 303 value = 33362.2 19->23 21 mae = 16292.941 samples = 45 value = 57882.167 20->21 22 mae = 10184.097 samples = 132 value = 46630.399 20->22 24 mae = 7464.802 samples = 125 value = 35729.8 23->24 25 mae = 6860.887 samples = 178 value = 30923.35 23->25 27 bldg_age <= 115.5 mae = 5005.107 samples = 91 value = 22113.3 26->27 30 3 bedrooms days vacant <= 6.5 mae = 2913.894 samples = 35 value = 26206.1 26->30 28 mae = 4773.313 samples = 78 value = 21679.05 27->28 29 mae = 4363.446 samples = 13 value = 27422.2 27->29 31 mae = 2536.153 samples = 32 value = 26302.65 30->31 32 mae = 113.767 samples = 3 value = 19232.2 30->32 34 basement sq ft <= 6926.5 mae = 13623.593 samples = 315 value = 92493.2 33->34 49 bldg_age <= 46.5 mae = 20370.539 samples = 437 value = 74399.998 33->49 35 bldg_age <= 95.5 mae = 12413.766 samples = 268 value = 89942.2 34->35 42 bldg_age <= 97.5 mae = 11388.481 samples = 47 value = 109016.001 34->42 36 month_february <= 0.5 mae = 11860.66 samples = 156 value = 93604.451 35->36 39 bldg_age <= 98.5 mae = 11351.429 samples = 112 value = 84475.6 35->39 37 mae = 11897.664 samples = 143 value = 94008.201 36->37 38 mae = 8436.323 samples = 13 value = 82130.301 36->38 40 mae = 9886.606 samples = 63 value = 86315.5 39->40 41 mae = 11506.579 samples = 49 value = 79307.799 39->41 43 month_december <= 0.5 mae = 14605.0 samples = 11 value = 98057.8 42->43 46 bldg_age <= 98.5 mae = 9164.739 samples = 36 value = 110125.999 42->46 44 mae = 10623.88 samples = 10 value = 96253.601 43->44 45 mae = 0.0 samples = 1 value = 152474.0 43->45 47 mae = 10313.283 samples = 12 value = 122149.5 46->47 48 mae = 6608.217 samples = 24 value = 109301.5 46->48 50 basement sq ft <= 3419.5 mae = 17874.419 samples = 236 value = 81544.4 49->50 57 bldg_age <= 99.5 mae = 19061.985 samples = 201 value = 60810.0 49->57 51 bldg_age <= 28.5 mae = 13670.003 samples = 149 value = 76630.401 50->51 54 bldg_age <= 42.5 mae = 21757.04 samples = 87 value = 91655.398 50->54 52 mae = 13585.646 samples = 67 value = 84970.801 51->52 53 mae = 11291.546 samples = 82 value = 70375.849 51->53 55 mae = 23096.475 samples = 61 value = 95437.202 54->55 56 mae = 14583.158 samples = 26 value = 81885.8 54->56 58 1 bedrooms days vacant <= 29.5 mae = 19282.228 samples = 175 value = 64348.899 57->58 61 0 bedrooms - days vacant <= 9.0 mae = 5194.788 samples = 26 value = 43764.2 57->61 59 mae = 17146.433 samples = 142 value = 61657.349 58->59 60 mae = 24541.612 samples = 33 value = 75707.301 58->60 62 mae = 6812.389 samples = 9 value = 47850.899 61->62 63 mae = 3373.894 samples = 17 value = 41563.501 61->63 65 total # bedrooms  in building <= 76.0 mae = 55644.755 samples = 958 value = 172613.0 64->65 96 low-income housing_no <= 0.5 mae = 58082.125 samples = 145 value = 404376.005 64->96 66 if multi-family sum of apt sq ft <= 30084.0 mae = 39981.277 samples = 464 value = 136590.528 65->66 81 if environmental certification: type_other (type in) <= 0.5 mae = 52692.777 samples = 494 value = 204993.998 65->81 67 hot water system fuel_natural gas <= 0.5 mae = 35614.382 samples = 328 value = 117267.5 66->67 74 bldg_age <= 9.5 mae = 31538.349 samples = 136 value = 173733.289 66->74 68 0 bedrooms - days vacant <= 110.0 mae = 6514.636 samples = 41 value = 82842.88 67->68 71 total # bedrooms  in building <= 55.5 mae = 35208.842 samples = 287 value = 125019.0 67->71 69 mae = 6325.649 samples = 36 value = 83549.26 68->69 70 mae = 2906.528 samples = 5 value = 73010.9 68->70 72 mae = 19611.665 samples = 34 value = 71566.6 71->72 73 mae = 33109.299 samples = 253 value = 130016.0 71->73 75 bldg_age <= 7.5 mae = 17953.643 samples = 28 value = 200932.998 74->75 78 0 bedrooms - days vacant <= 588.5 mae = 29608.847 samples = 108 value = 158947.433 74->78 76 mae = 16848.2 samples = 10 value = 213088.998 75->76 77 mae = 13425.223 samples = 18 value = 190414.998 75->77 79 mae = 24721.435 samples = 93 value = 155281.001 78->79 80 mae = 34588.094 samples = 15 value = 216183.208 78->80 82 # stories <= 6.0 mae = 38607.548 samples = 357 value = 190463.001 81->82 89 cooling system_ducted central ac (outdoor condenser)  <= 0.5 mae = 60681.861 samples = 137 value = 272226.002 81->89 83 heating system_boiler (hot water) <= 0.5 mae = 35822.892 samples = 268 value = 201774.497 82->83 86 bldg_age <= 41.5 mae = 22715.0 samples = 89 value = 140531.002 82->86 84 mae = 23459.581 samples = 142 value = 188315.498 83->84 85 mae = 42049.23 samples = 126 value = 222371.0 83->85 87 mae = 17731.986 samples = 73 value = 135988.001 86->87 88 mae = 14012.814 samples = 16 value = 189486.502 86->88 90 total # bedrooms  in building <= 94.0 mae = 36507.831 samples = 89 value = 237019.998 89->90 93 bldg_age <= 5.5 mae = 46313.375 samples = 48 value = 350541.5 89->93 91 mae = 22834.907 samples = 43 value = 202448.0 90->91 92 mae = 22757.869 samples = 46 value = 267529.497 90->92 94 mae = 41105.7 samples = 10 value = 391923.5 93->94 95 mae = 37652.447 samples = 38 value = 331587.0 93->95 97 month_may <= 0.5 mae = 25526.98 samples = 51 value = 468904.997 96->97 104 # of 1 bedrooms * days in month <= 2242.0 mae = 42569.862 samples = 94 value = 360938.505 96->104 98 # of 2 bedrooms * days in month <= 1254.0 mae = 23129.934 samples = 46 value = 465587.006 97->98 103 mae = 14909.602 samples = 5 value = 509286.008 97->103 99 mae = 7779.499 samples = 4 value = 426174.498 98->99 100 month_july <= 0.5 mae = 21440.499 samples = 42 value = 468188.998 98->100 101 mae = 20862.21 samples = 38 value = 465587.006 100->101 102 mae = 10312.75 samples = 4 value = 496027.497 100->102 105 0 bedrooms - days vacant <= 0.5 mae = 31108.482 samples = 54 value = 344555.998 104->105 112 bldg_age <= 10.5 mae = 46671.425 samples = 40 value = 400047.502 104->112 106 1 bedrooms days vacant <= 0.5 mae = 28564.109 samples = 46 value = 337522.003 105->106 109 month_november <= 0.5 mae = 27289.376 samples = 8 value = 393008.005 105->109 107 mae = 20658.068 samples = 15 value = 351304.005 106->107 108 mae = 28389.259 samples = 31 value = 327960.997 106->108 110 mae = 22207.571 samples = 7 value = 405409.004 109->110 111 mae = 0.0 samples = 1 value = 342546.997 109->111 113 month_august <= 0.5 mae = 35728.666 samples = 9 value = 325955.995 112->113 116 month_august <= 0.5 mae = 40605.581 samples = 31 value = 404376.005 112->116 114 mae = 22755.998 samples = 8 value = 325661.501 113->114 115 mae = 0.0 samples = 1 value = 465466.003 113->115 117 mae = 35814.357 samples = 28 value = 403158.499 116->117 118 mae = 19794.998 samples = 3 value = 492894.0 116->118

In [88]:
clf_ = DecisionTreeRegressor(criterion='mae')
clf2 = train_model(X=X, y=y, estimator=clf_, params=[{'max_depth': [None], 'min_samples_split': [4, 6], 'min_samples_leaf':[4]}])
makeplot(estimator=clf2, X=X, y=y, X_valid=X_valid, y_valid=y_valid, title='Decision Tree Regression with 10-Fold Cross-Validation')
plot_variables(features=features, data=clf2.best_estimator_.feature_importances_, X=X)


The model was trained with 83 features.  Of those 83 features contribute to the outcome.

In [89]:
clfplot=export_graphviz(clf2.best_estimator_, out_file='my_tree2.dot', feature_names=features)
with open('my_tree2.dot') as f:
    dot_graph=f.read()
graphviz.Source(dot_graph)


Out[89]:
Tree 0 total # bedrooms  in building <= 53.0 mae = 72424.854 samples = 3364 value = 69099.949 1 total # bedrooms  in building <= 24.0 mae = 28218.987 samples = 2261 value = 31225.2 0->1 True 814 total # bedrooms  in building <= 126.5 mae = 77549.295 samples = 1103 value = 185532.998 0->814 False 2 total # bedrooms  in building <= 8.5 mae = 12237.745 samples = 1509 value = 18835.5 1->2 521 if multi-family, # units <= 25.5 mae = 19269.76 samples = 752 value = 83264.15 1->521 3 if multi-family, # units <= 1.0 mae = 5861.254 samples = 903 value = 13479.5 2->3 314 single family building type_shared wall  <= 0.5 mae = 11378.372 samples = 606 value = 34499.352 2->314 4 total # bedrooms  in building <= 5.0 mae = 3370.972 samples = 223 value = 4163.98 3->4 77 # stories <= 2.25 mae = 4785.348 samples = 680 value = 15622.8 3->77 5 # of 3 bedrooms * days in month <= 14.0 mae = 2321.16 samples = 176 value = 3189.86 4->5 62 bldg_age <= 23.5 mae = 1627.534 samples = 47 value = 10400.2 4->62 6 gross bldg sq ft <= 1426.5 mae = 1070.696 samples = 93 value = 2075.1 5->6 33 bldg_age <= 114.5 mae = 2623.43 samples = 83 value = 5723.81 5->33 7 1 bedrooms days vacant <= 10.5 mae = 589.135 samples = 50 value = 1492.785 6->7 22 bldg_age <= 114.5 mae = 1197.26 samples = 43 value = 3189.86 6->22 8 bldg_age <= 101.5 mae = 513.788 samples = 45 value = 1502.43 7->8 21 mae = 190.137 samples = 5 value = 199.467 7->21 9 bldg_age <= 98.5 mae = 481.172 samples = 38 value = 1586.08 8->9 20 mae = 288.005 samples = 7 value = 751.217 8->20 10 mae = 167.652 samples = 9 value = 1489.57 9->10 11 bldg_age <= 100.5 mae = 530.957 samples = 29 value = 2127.52 9->11 12 # of 1 bedrooms * days in month <= 30.5 mae = 381.571 samples = 18 value = 2153.92 11->12 17 # of days in month <= 30.5 mae = 467.227 samples = 11 value = 1402.7 11->17 13 mae = 266.076 samples = 7 value = 1586.08 12->13 14 bldg_age <= 99.5 mae = 298.907 samples = 11 value = 2168.17 12->14 15 mae = 450.73 samples = 5 value = 2340.52 14->15 16 mae = 110.42 samples = 6 value = 2153.92 14->16 18 mae = 555.158 samples = 5 value = 1296.53 17->18 19 mae = 376.256 samples = 6 value = 1402.7 17->19 23 bldg_age <= 113.5 mae = 740.822 samples = 17 value = 3450.45 22->23 26 bldg_age <= 116.5 mae = 1222.91 samples = 26 value = 2199.57 22->26 24 mae = 425.498 samples = 10 value = 3216.4 23->24 25 mae = 965.851 samples = 7 value = 4198.45 23->25 27 bldg_age <= 115.5 mae = 812.48 samples = 21 value = 2075.1 26->27 32 mae = 1600.482 samples = 5 value = 5453.16 26->32 28 mae = 909.044 samples = 10 value = 2625.585 27->28 29 # of days in month <= 30.5 mae = 585.936 samples = 11 value = 1476.7 27->29 30 mae = 436.818 samples = 5 value = 2055.79 29->30 31 mae = 526.818 samples = 6 value = 1211.275 29->31 34 bldg_age <= 10.5 mae = 2595.497 samples = 56 value = 4231.2 33->34 53 # of 3 bedrooms * days in month <= 30.5 mae = 1573.461 samples = 27 value = 6901.52 33->53 35 mae = 3785.83 samples = 5 value = 11655.3 34->35 36 month_august <= 0.5 mae = 1870.714 samples = 51 value = 3947.51 34->36 37 month_january <= 0.5 mae = 1792.367 samples = 47 value = 3757.24 36->37 52 mae = 1496.25 samples = 4 value = 6913.16 36->52 38 single family building type_detached  <= 0.5 mae = 1515.483 samples = 42 value = 3691.74 37->38 51 mae = 3064.706 samples = 5 value = 6853.03 37->51 39 bldg_age <= 12.5 mae = 1091.775 samples = 27 value = 3489.06 38->39 48 bldg_age <= 113.5 mae = 2004.972 samples = 15 value = 4405.96 38->48 40 bldg_age <= 11.5 mae = 860.991 samples = 15 value = 3691.74 39->40 45 bldg_age <= 13.5 mae = 1121.269 samples = 12 value = 2826.51 39->45 41 # of 3 bedrooms * days in month <= 30.5 mae = 1007.465 samples = 8 value = 3724.49 40->41 44 mae = 664.639 samples = 7 value = 3489.06 40->44 42 mae = 409.505 samples = 4 value = 3691.74 41->42 43 mae = 1572.675 samples = 4 value = 4231.2 41->43 46 mae = 1305.143 samples = 7 value = 2731.41 45->46 47 mae = 710.22 samples = 5 value = 3242.94 45->47 49 mae = 1670.745 samples = 8 value = 5221.38 48->49 50 mae = 2177.664 samples = 7 value = 3947.51 48->50 54 bldg_age <= 115.5 mae = 1170.835 samples = 11 value = 6327.02 53->54 57 bldg_age <= 115.5 mae = 1594.792 samples = 16 value = 7674.04 53->57 55 mae = 1714.975 samples = 4 value = 7369.695 54->55 56 mae = 818.534 samples = 7 value = 6037.47 54->56 58 mae = 2220.556 samples = 5 value = 8518.55 57->58 59 bldg_age <= 116.5 mae = 1224.807 samples = 11 value = 7577.53 57->59 60 mae = 1374.077 samples = 6 value = 7806.555 59->60 61 mae = 1045.684 samples = 5 value = 7577.53 59->61 63 # of 3 bedrooms * days in month <= 61.0 mae = 1859.206 samples = 19 value = 13040.3 62->63 70 # of 3 bedrooms * days in month <= 61.0 mae = 941.73 samples = 28 value = 9869.77 62->70 64 bldg_age <= 22.5 mae = 1829.209 samples = 9 value = 12104.1 63->64 67 bldg_age <= 22.5 mae = 1792.584 samples = 10 value = 13100.65 63->67 65 mae = 1973.76 samples = 5 value = 12104.1 64->65 66 mae = 1648.52 samples = 4 value = 12179.95 64->66 68 mae = 1422.64 samples = 5 value = 13161.0 67->68 69 mae = 2138.388 samples = 5 value = 13040.3 67->69 71 bldg_age <= 24.5 mae = 713.157 samples = 12 value = 9585.04 70->71 74 bldg_age <= 24.5 mae = 994.317 samples = 16 value = 10354.55 70->74 72 mae = 431.902 samples = 4 value = 9440.265 71->72 73 mae = 846.544 samples = 8 value = 9744.29 71->73 75 mae = 1078.213 samples = 7 value = 9990.46 74->75 76 mae = 883.538 samples = 9 value = 10400.2 74->76 78 gross bldg sq ft <= 3079.5 mae = 4579.459 samples = 632 value = 16099.9 77->78 295 4 bedrooms days vacant <= 1.0 mae = 859.804 samples = 48 value = 8412.37 77->295 79 bldg_age <= 13.5 mae = 5290.468 samples = 130 value = 11430.95 78->79 126 hot water system_no <= 0.5 mae = 3908.014 samples = 502 value = 16903.35 78->126 80 bldg_age <= 10.5 mae = 2544.309 samples = 34 value = 13834.2 79->80 93 bldg_age <= 116.5 mae = 5685.316 samples = 96 value = 10102.8 79->93 81 mae = 782.286 samples = 7 value = 11655.3 80->81 82 # of 3 bedrooms * days in month <= 61.0 mae = 2667.741 samples = 27 value = 14854.8 80->82 83 bldg_age <= 11.5 mae = 2129.35 samples = 12 value = 14044.1 82->83 88 bldg_age <= 12.5 mae = 2985.213 samples = 15 value = 14961.0 82->88 84 mae = 2740.2 samples = 4 value = 15006.85 83->84 85 bldg_age <= 12.5 mae = 1823.925 samples = 8 value = 14044.1 83->85 86 mae = 724.225 samples = 4 value = 14044.1 85->86 87 mae = 2923.625 samples = 4 value = 15347.05 85->87 89 bldg_age <= 11.5 mae = 2990.991 samples = 11 value = 14854.8 88->89 92 mae = 2942.775 samples = 4 value = 15805.5 88->92 90 mae = 2851.675 samples = 4 value = 14249.9 89->90 91 mae = 3070.6 samples = 7 value = 14854.8 89->91 94 2 bedrooms days vacant <= 1.5 mae = 3142.77 samples = 83 value = 9266.55 93->94 123 # of 3 bedrooms * days in month <= 14.0 mae = 17595.633 samples = 13 value = 25071.8 93->123 95 if multi-family sum of apt sq ft <= 2105.0 mae = 3079.06 samples = 76 value = 9334.1 94->95 122 mae = 2620.776 samples = 7 value = 5322.86 94->122 96 bldg_age <= 115.5 mae = 4093.456 samples = 30 value = 10448.85 95->96 107 bldg_age <= 115.5 mae = 2247.526 samples = 46 value = 8692.075 95->107 97 bldg_age <= 113.5 mae = 3925.895 samples = 19 value = 11944.9 96->97 104 # of 2 bedrooms * days in month <= 61.0 mae = 2980.081 samples = 11 value = 7876.73 96->104 98 mae = 5584.417 samples = 6 value = 18315.25 97->98 99 # of days in month <= 30.5 mae = 2617.708 samples = 13 value = 11751.8 97->99 100 mae = 1735.573 samples = 4 value = 9440.265 99->100 101 bldg_age <= 114.5 mae = 2510.029 samples = 9 value = 11944.9 99->101 102 mae = 2057.005 samples = 4 value = 11848.35 101->102 103 mae = 2793.108 samples = 5 value = 12341.6 101->103 105 mae = 4082.998 samples = 5 value = 7833.27 104->105 106 mae = 2053.74 samples = 6 value = 7968.415 104->106 108 bldg_age <= 113.5 mae = 1527.432 samples = 35 value = 8111.84 107->108 119 # of 2 bedrooms * days in month <= 30.5 mae = 2269.667 samples = 11 value = 12116.3 107->119 109 # of days in month <= 30.5 mae = 1470.333 samples = 13 value = 7065.37 108->109 114 bldg_age <= 114.5 mae = 1420.311 samples = 22 value = 8383.225 108->114 110 mae = 632.105 samples = 4 value = 6190.48 109->110 111 gross bldg sq ft <= 2861.0 mae = 1554.32 samples = 9 value = 7736.38 109->111 112 mae = 2319.837 samples = 4 value = 8958.525 111->112 113 mae = 941.906 samples = 5 value = 7736.38 111->113 115 # of days in month <= 30.5 mae = 1239.045 samples = 11 value = 9387.19 114->115 118 mae = 1090.957 samples = 11 value = 7543.34 114->118 116 mae = 1231.93 samples = 4 value = 9305.11 115->116 117 mae = 1243.11 samples = 7 value = 9387.19 115->117 120 mae = 2385.408 samples = 5 value = 11944.9 119->120 121 mae = 2144.65 samples = 6 value = 13080.2 119->121 124 mae = 2816.89 samples = 7 value = 14319.2 123->124 125 mae = 4841.467 samples = 6 value = 52095.1 123->125 127 cooling system_ducted central ac (outdoor condenser)  <= 0.5 mae = 3895.564 samples = 461 value = 17243.6 126->127 286 bldg_age <= 115.5 mae = 2364.221 samples = 41 value = 13658.0 126->286 128 bldg_age <= 114.5 mae = 4049.082 samples = 339 value = 16457.0 127->128 243 bldg_age <= 114.5 mae = 2927.827 samples = 122 value = 19215.15 127->243 129 total # bedrooms  in building <= 7.0 mae = 3962.983 samples = 288 value = 16001.75 128->129 226 bldg_age <= 116.5 mae = 3550.716 samples = 51 value = 18923.0 128->226 130 bldg_age <= 113.5 mae = 3160.145 samples = 84 value = 11560.5 129->130 159 bldg_age <= 20.5 mae = 3415.426 samples = 204 value = 17229.95 129->159 131 3 bedrooms days vacant <= 0.5 mae = 3083.995 samples = 67 value = 10699.2 130->131 154 2 bedrooms days vacant <= 0.5 mae = 2493.072 samples = 17 value = 13551.9 130->154 132 bldg_age <= 22.5 mae = 2967.374 samples = 61 value = 11365.8 131->132 153 mae = 3460.48 samples = 6 value = 7547.865 131->153 133 bldg_age <= 21.5 mae = 3213.87 samples = 30 value = 10405.45 132->133 142 basement sq ft <= 751.0 mae = 2495.833 samples = 31 value = 12327.5 132->142 134 bldg_age <= 20.5 mae = 4034.665 samples = 19 value = 10757.7 133->134 139 # of 3 bedrooms * days in month <= 61.0 mae = 1714.353 samples = 11 value = 10128.3 133->139 135 # of days in month <= 30.5 mae = 1346.207 samples = 12 value = 10533.3 134->135 138 mae = 7856.15 samples = 7 value = 16268.8 134->138 136 mae = 1360.062 samples = 5 value = 10757.7 135->136 137 mae = 1272.196 samples = 7 value = 10308.9 135->137 140 mae = 1856.694 samples = 5 value = 9700.86 139->140 141 mae = 1524.495 samples = 6 value = 10368.2 139->141 143 # of 2 bedrooms * days in month <= 61.0 mae = 2112.149 samples = 11 value = 9435.45 142->143 146 # of days in month <= 30.5 mae = 2168.347 samples = 20 value = 13042.0 142->146 144 mae = 1839.513 samples = 4 value = 8584.36 143->144 145 mae = 2050.969 samples = 7 value = 9487.2 143->145 147 mae = 2106.383 samples = 7 value = 12552.9 146->147 148 bldg_age <= 23.5 mae = 2125.981 samples = 13 value = 13537.4 146->148 149 mae = 1768.208 samples = 4 value = 13129.6 148->149 150 single family building type_detached  <= 0.5 mae = 2281.924 samples = 9 value = 13565.0 148->150 151 mae = 1954.444 samples = 5 value = 13565.0 150->151 152 mae = 2691.275 samples = 4 value = 14223.35 150->152 155 single family building type_detached  <= 0.5 mae = 1623.32 samples = 13 value = 13638.7 154->155 158 mae = 3630.765 samples = 4 value = 8221.595 154->158 156 mae = 1515.294 samples = 7 value = 13291.3 155->156 157 mae = 1148.217 samples = 6 value = 15358.45 155->157 160 4 bedrooms days vacant <= 2.0 mae = 3658.883 samples = 112 value = 17801.2 159->160 193 month_july <= 0.5 mae = 2914.129 samples = 92 value = 16255.2 159->193 161 # of 4 bedrooms * days in month <= 61.0 mae = 3610.485 samples = 108 value = 18035.05 160->161 192 mae = 1257.475 samples = 4 value = 12622.9 160->192 162 bldg_age <= 6.5 mae = 2853.509 samples = 68 value = 17612.8 161->162 181 month_august <= 0.5 mae = 4605.3 samples = 40 value = 20250.75 161->181 163 mae = 3015.4 samples = 7 value = 12910.0 162->163 164 month_june <= 0.5 mae = 2757.005 samples = 61 value = 17663.5 162->164 165 month_july <= 0.5 mae = 2822.386 samples = 51 value = 17562.1 164->165 180 mae = 2222.0 samples = 10 value = 19358.1 164->180 166 month_february <= 0.5 mae = 2874.157 samples = 47 value = 17460.8 165->166 179 mae = 1828.4 samples = 4 value = 18907.75 165->179 167 month_november <= 0.5 mae = 2362.495 samples = 40 value = 17511.45 166->167 178 mae = 5659.457 samples = 7 value = 16491.4 166->178 168 month_september <= 0.5 mae = 2264.15 samples = 32 value = 17718.65 167->168 175 bldg_age <= 19.5 mae = 2640.05 samples = 8 value = 17077.1 167->175 169 bldg_age <= 7.5 mae = 2114.864 samples = 22 value = 17612.8 168->169 174 mae = 2560.66 samples = 10 value = 18018.55 168->174 170 mae = 1147.84 samples = 5 value = 18049.5 169->170 171 # of 2 bedrooms * days in month <= 30.5 mae = 2370.612 samples = 17 value = 17562.1 169->171 172 mae = 1876.912 samples = 8 value = 17670.7 171->172 173 mae = 2775.322 samples = 9 value = 17254.9 171->173 176 mae = 2887.7 samples = 4 value = 17156.75 175->176 177 mae = 2392.4 samples = 4 value = 15600.4 175->177 182 month_october <= 0.5 mae = 4315.468 samples = 34 value = 18643.1 181->182 191 mae = 3305.183 samples = 6 value = 24420.6 181->191 183 month_december <= 0.5 mae = 4242.732 samples = 28 value = 19797.2 182->183 190 mae = 3060.533 samples = 6 value = 15532.85 182->190 184 month_may <= 0.5 mae = 4361.468 samples = 22 value = 21075.6 183->184 189 mae = 2698.6 samples = 6 value = 17875.8 183->189 185 month_january <= 0.5 mae = 4158.4 samples = 17 value = 21355.2 184->185 188 mae = 4389.8 samples = 5 value = 18044.7 184->188 186 mae = 3717.242 samples = 12 value = 21075.6 185->186 187 mae = 4951.68 samples = 5 value = 22682.7 185->187 194 # of 4 bedrooms * days in month <= 59.0 mae = 2538.902 samples = 83 value = 15853.8 193->194 223 bldg_age <= 22.5 mae = 4359.611 samples = 9 value = 21157.7 193->223 195 mae = 810.029 samples = 7 value = 14340.5 194->195 196 bldg_age <= 21.5 mae = 2586.632 samples = 76 value = 16255.2 194->196 197 month_august <= 0.5 mae = 2009.733 samples = 33 value = 15453.2 196->197 210 month_may <= 0.5 mae = 2857.251 samples = 43 value = 17182.6 196->210 198 month_december <= 0.5 mae = 1438.862 samples = 29 value = 15173.3 197->198 209 mae = 5001.025 samples = 4 value = 20161.0 197->209 199 # of days in month <= 30.5 mae = 1192.88 samples = 25 value = 15033.4 198->199 208 mae = 1639.725 samples = 4 value = 18522.75 198->208 200 month_september <= 0.5 mae = 1234.518 samples = 11 value = 15795.9 199->200 203 month_march <= 0.5 mae = 1105.7 samples = 14 value = 14973.05 199->203 201 mae = 1276.071 samples = 7 value = 15853.8 200->201 202 mae = 1147.325 samples = 4 value = 15122.7 200->202 204 month_january <= 0.5 mae = 1190.52 samples = 10 value = 14973.05 203->204 207 mae = 893.65 samples = 4 value = 14733.55 203->207 205 mae = 1598.133 samples = 6 value = 14854.8 204->205 206 mae = 579.1 samples = 4 value = 14973.05 204->206 211 month_january <= 0.5 mae = 2868.016 samples = 37 value = 16794.8 210->211 222 mae = 2286.033 samples = 6 value = 18582.8 210->222 212 month_august <= 0.5 mae = 3038.287 samples = 30 value = 16359.65 211->212 221 mae = 1547.829 samples = 7 value = 18025.8 211->221 213 month_march <= 0.5 mae = 2240.644 samples = 25 value = 16265.7 212->213 220 mae = 6804.86 samples = 5 value = 17373.9 212->220 214 bldg_age <= 22.5 mae = 2728.237 samples = 19 value = 16085.4 213->214 219 mae = 666.55 samples = 6 value = 16359.65 213->219 215 month_november <= 0.5 mae = 2727.869 samples = 13 value = 15795.9 214->215 218 mae = 2680.783 samples = 6 value = 17127.75 214->218 216 mae = 2437.722 samples = 9 value = 15795.9 215->216 217 mae = 3380.7 samples = 4 value = 16070.95 215->217 224 mae = 4350.32 samples = 5 value = 21157.7 223->224 225 mae = 4371.225 samples = 4 value = 20298.6 223->225 227 month_august <= 0.5 mae = 3111.149 samples = 39 value = 18204.0 226->227 238 # of 3 bedrooms * days in month <= 58.0 mae = 4093.425 samples = 12 value = 21449.8 226->238 228 # of 2 bedrooms * days in month <= 28.0 mae = 3095.751 samples = 35 value = 18111.1 227->228 237 mae = 2600.15 samples = 4 value = 19617.9 227->237 229 # of 3 bedrooms * days in month <= 61.0 mae = 3517.733 samples = 18 value = 20238.35 228->229 234 bldg_age <= 115.5 mae = 2413.253 samples = 17 value = 17330.5 228->234 230 bldg_age <= 115.5 mae = 2988.1 samples = 8 value = 22665.9 229->230 233 mae = 3223.28 samples = 10 value = 18563.5 229->233 231 mae = 1809.875 samples = 4 value = 22999.15 230->231 232 mae = 4014.225 samples = 4 value = 19733.9 230->232 235 mae = 2308.644 samples = 9 value = 16085.4 234->235 236 mae = 2180.15 samples = 8 value = 18430.15 234->236 239 # of days in month <= 30.5 mae = 1648.712 samples = 8 value = 20436.05 238->239 242 mae = 5696.15 samples = 4 value = 30887.5 238->242 240 mae = 1014.4 samples = 4 value = 19369.0 239->240 241 mae = 1653.075 samples = 4 value = 21449.8 239->241 244 gross bldg sq ft <= 4552.5 mae = 2826.576 samples = 94 value = 20011.85 243->244 277 basement sq ft <= 2240.0 mae = 2404.157 samples = 28 value = 16905.8 243->277 245 bldg_age <= 17.5 mae = 2418.452 samples = 42 value = 18423.55 244->245 260 3 bedrooms days vacant <= 0.5 mae = 2695.913 samples = 52 value = 21136.75 244->260 246 bldg_age <= 16.5 mae = 2389.481 samples = 36 value = 18715.5 245->246 259 mae = 820.517 samples = 6 value = 15773.15 245->259 247 # of days in month <= 30.5 mae = 2691.424 samples = 25 value = 18647.9 246->247 256 # of days in month <= 30.5 mae = 1588.736 samples = 11 value = 19907.5 246->256 248 bldg_age <= 14.5 mae = 2976.655 samples = 11 value = 18300.4 247->248 251 bldg_age <= 14.5 mae = 2423.179 samples = 14 value = 18942.35 247->251 249 mae = 1259.65 samples = 4 value = 18141.15 248->249 250 mae = 3908.157 samples = 7 value = 18647.9 248->250 252 mae = 1227.82 samples = 5 value = 18783.1 251->252 253 bldg_age <= 15.5 mae = 3051.878 samples = 9 value = 19101.6 251->253 254 mae = 3438.22 samples = 5 value = 22113.3 253->254 255 mae = 1816.025 samples = 4 value = 18722.75 253->255 257 mae = 1853.2 samples = 4 value = 18879.75 256->257 258 mae = 1437.614 samples = 7 value = 19907.5 256->258 261 bldg_age <= 113.5 mae = 2580.955 samples = 47 value = 21414.1 260->261 276 mae = 2731.38 samples = 5 value = 18073.6 260->276 262 # of 3 bedrooms * days in month <= 61.0 mae = 2948.785 samples = 33 value = 21804.4 261->262 271 # of 2 bedrooms * days in month <= 30.5 mae = 1351.093 samples = 14 value = 20416.95 261->271 263 bldg_age <= 112.5 mae = 2200.362 samples = 13 value = 21414.1 262->263 266 basement sq ft <= 2240.0 mae = 3275.235 samples = 20 value = 22591.4 262->266 264 mae = 2855.114 samples = 7 value = 21804.4 263->264 265 mae = 1371.433 samples = 6 value = 21172.9 263->265 267 mae = 1234.717 samples = 6 value = 23744.95 266->267 268 bldg_age <= 111.5 mae = 3985.671 samples = 14 value = 22292.2 266->268 269 mae = 2891.48 samples = 5 value = 22012.1 268->269 270 mae = 4527.067 samples = 9 value = 22610.5 268->270 272 mae = 1034.683 samples = 6 value = 20595.3 271->272 273 basement sq ft <= 2240.0 mae = 1538.925 samples = 8 value = 20298.75 271->273 274 mae = 1048.3 samples = 4 value = 20298.75 273->274 275 mae = 2029.55 samples = 4 value = 20436.2 273->275 278 bldg_age <= 115.5 mae = 2280.114 samples = 21 value = 15829.6 277->278 285 mae = 929.386 samples = 7 value = 19328.3 277->285 279 mae = 1359.15 samples = 6 value = 12596.7 278->279 280 bldg_age <= 116.5 mae = 1830.04 samples = 15 value = 16698.3 278->280 281 # of 2 bedrooms * days in month <= 30.5 mae = 1531.011 samples = 9 value = 17542.8 280->281 284 mae = 1761.433 samples = 6 value = 16092.65 280->284 282 mae = 1631.25 samples = 4 value = 16505.25 281->282 283 mae = 1277.06 samples = 5 value = 17813.2 281->283 287 bldg_age <= 113.5 mae = 1541.952 samples = 28 value = 13539.8 286->287 294 mae = 3849.892 samples = 13 value = 15665.6 286->294 288 mae = 1374.657 samples = 7 value = 13937.9 287->288 289 # of 4 bedrooms * days in month <= 30.5 mae = 1528.779 samples = 21 value = 12972.8 287->289 290 mae = 1432.305 samples = 8 value = 11781.45 289->290 291 bldg_age <= 114.5 mae = 1310.686 samples = 13 value = 13658.0 289->291 292 mae = 1280.029 samples = 7 value = 13498.8 291->292 293 mae = 1319.92 samples = 6 value = 13807.6 291->293 296 month_june <= 0.5 mae = 727.24 samples = 43 value = 8571.62 295->296 313 mae = 412.592 samples = 5 value = 6604.85 295->313 297 month_july <= 0.5 mae = 623.009 samples = 39 value = 8475.13 296->297 312 mae = 1140.268 samples = 4 value = 10099.005 296->312 298 month_august <= 0.5 mae = 613.698 samples = 35 value = 8412.37 297->298 311 mae = 348.565 samples = 4 value = 9022.66 297->311 299 month_march <= 0.5 mae = 562.36 samples = 30 value = 8265.895 298->299 310 mae = 588.74 samples = 5 value = 9266.55 298->310 300 month_september <= 0.5 mae = 572.678 samples = 26 value = 8202.465 299->300 309 mae = 255.702 samples = 4 value = 8807.655 299->309 301 bldg_age <= 32.5 mae = 608.585 samples = 22 value = 8175.93 300->301 308 mae = 256.975 samples = 4 value = 8557.145 300->308 302 bldg_age <= 31.5 mae = 285.341 samples = 10 value = 8202.465 301->302 305 bldg_age <= 33.5 mae = 842.561 samples = 12 value = 7770.55 301->305 303 mae = 341.868 samples = 5 value = 7992.14 302->303 304 mae = 170.828 samples = 5 value = 8282.07 302->304 306 mae = 1118.058 samples = 5 value = 7481.0 305->306 307 mae = 576.839 samples = 7 value = 7963.57 305->307 315 cooling system_window ac  <= 0.5 mae = 11227.654 samples = 480 value = 36983.65 314->315 480 # of 2 bedrooms * days in month <= 59.0 mae = 4763.505 samples = 126 value = 24099.65 314->480 316 bldg_age <= 81.0 mae = 12287.923 samples = 177 value = 48184.8 315->316 377 single family building type_detached  <= 0.5 mae = 7579.429 samples = 303 value = 33362.2 315->377 317 bldg_age <= 48.5 mae = 10184.097 samples = 132 value = 46630.399 316->317 364 bldg_age <= 115.5 mae = 16292.941 samples = 45 value = 57882.167 316->364 318 # of 1 bedrooms * days in month <= 598.5 mae = 10318.766 samples = 120 value = 47587.001 317->318 359 1 bedrooms days vacant <= 19.5 mae = 4261.225 samples = 12 value = 39087.0 317->359 319 bldg_age <= 7.5 mae = 12163.116 samples = 50 value = 41179.95 318->319 336 bldg_age <= 45.5 mae = 8278.464 samples = 70 value = 48722.851 318->336 320 2 bedrooms days vacant <= 31.5 mae = 18430.211 samples = 18 value = 59077.449 319->320 327 2 bedrooms days vacant <= 0.5 mae = 5708.719 samples = 32 value = 39325.799 319->327 321 1 bedrooms days vacant <= 0.5 mae = 16935.043 samples = 14 value = 68132.9 320->321 326 mae = 5169.8 samples = 4 value = 42986.549 320->326 322 # of 1 bedrooms * days in month <= 152.5 mae = 16521.55 samples = 10 value = 74602.649 321->322 325 mae = 6224.374 samples = 4 value = 50935.5 321->325 323 mae = 6534.9 samples = 4 value = 68132.9 322->323 324 mae = 18613.983 samples = 6 value = 78813.7 322->324 328 # of 0 bedrooms * days in month <= 177.0 mae = 6326.616 samples = 25 value = 40993.399 327->328 335 mae = 2318.515 samples = 7 value = 37697.2 327->335 329 mae = 7389.912 samples = 8 value = 39167.6 328->329 330 bldg_age <= 8.5 mae = 5416.283 samples = 17 value = 41131.7 328->330 331 mae = 7112.08 samples = 5 value = 42786.801 330->331 332 # of 0 bedrooms * days in month <= 183.0 mae = 4571.775 samples = 12 value = 41102.7 330->332 333 mae = 1716.16 samples = 5 value = 41131.7 332->333 334 mae = 6603.215 samples = 7 value = 41073.699 332->334 337 1 bedrooms days vacant <= 8.0 mae = 11050.894 samples = 16 value = 56092.35 336->337 344 bldg_age <= 47.5 mae = 6992.037 samples = 54 value = 47239.1 336->344 338 # of days in month <= 30.5 mae = 11574.208 samples = 12 value = 65869.55 337->338 343 mae = 3932.05 samples = 4 value = 49722.799 337->343 339 mae = 7838.025 samples = 4 value = 65869.55 338->339 340 1 bedrooms days vacant <= 0.5 mae = 13442.3 samples = 8 value = 61831.5 338->340 341 mae = 15765.024 samples = 4 value = 62361.55 340->341 342 mae = 11119.575 samples = 4 value = 61831.5 340->342 345 1 bedrooms days vacant <= 27.0 mae = 8919.782 samples = 34 value = 44076.05 344->345 354 1 bedrooms days vacant <= 0.5 mae = 2973.56 samples = 20 value = 48184.8 344->354 346 1 bedrooms days vacant <= 0.5 mae = 8581.365 samples = 29 value = 45826.099 345->346 353 mae = 5071.74 samples = 5 value = 34032.6 345->353 347 # of days in month <= 30.5 mae = 6155.689 samples = 18 value = 43097.799 346->347 350 1 bedrooms days vacant <= 3.0 mae = 10143.518 samples = 11 value = 50217.401 346->350 348 mae = 5835.983 samples = 6 value = 42263.049 347->348 349 mae = 6080.041 samples = 12 value = 43804.299 347->349 351 mae = 11219.56 samples = 5 value = 54782.599 350->351 352 mae = 7808.417 samples = 6 value = 47407.6 350->352 355 mae = 2516.3 samples = 10 value = 50543.45 354->355 356 # of 1 bedrooms * days in month <= 640.5 mae = 2437.34 samples = 10 value = 46467.35 354->356 357 mae = 2202.875 samples = 4 value = 45485.5 356->357 358 mae = 2079.15 samples = 6 value = 48016.3 356->358 360 mae = 5808.676 samples = 4 value = 44822.8 359->360 361 1 bedrooms days vacant <= 47.0 mae = 2657.625 samples = 8 value = 38152.2 359->361 362 mae = 2023.925 samples = 4 value = 33584.25 361->362 363 mae = 489.125 samples = 4 value = 39423.95 361->363 365 bldg_age <= 113.5 mae = 14459.456 samples = 27 value = 69301.409 364->365 372 3 bedrooms days vacant <= 15.0 mae = 12587.054 samples = 18 value = 46782.888 364->372 366 mae = 12151.716 samples = 9 value = 81251.82 365->366 367 bldg_age <= 114.5 mae = 13745.234 samples = 18 value = 63681.086 365->367 368 # of days in month <= 30.5 mae = 13867.635 samples = 11 value = 57982.655 367->368 371 mae = 9662.69 samples = 7 value = 72221.912 367->371 369 mae = 16148.167 samples = 4 value = 48616.45 368->369 370 mae = 10238.791 samples = 7 value = 60827.888 368->370 373 bldg_age <= 116.5 mae = 11885.946 samples = 13 value = 51303.233 372->373 376 mae = 5480.447 samples = 5 value = 32716.603 372->376 374 mae = 13043.348 samples = 6 value = 57502.95 373->374 375 mae = 7968.516 samples = 7 value = 46150.484 373->375 378 bldg_age <= 98.5 mae = 7464.802 samples = 125 value = 35729.8 377->378 419 bldg_age <= 74.5 mae = 6860.887 samples = 178 value = 30923.35 377->419 379 1 bedrooms days vacant <= 52.5 mae = 6673.091 samples = 57 value = 38554.2 378->379 398 month_may <= 0.5 mae = 7541.465 samples = 68 value = 34419.3 378->398 380 1 bedrooms days vacant <= 0.5 mae = 5578.408 samples = 53 value = 38767.5 379->380 397 mae = 13531.275 samples = 4 value = 22301.5 379->397 381 bldg_age <= 19.5 mae = 6370.794 samples = 32 value = 40959.65 380->381 392 1 bedrooms days vacant <= 30.5 mae = 3467.791 samples = 21 value = 36123.5 380->392 382 mae = 14632.4 samples = 5 value = 43780.0 381->382 383 bldg_age <= 21.5 mae = 4502.585 samples = 27 value = 40274.401 381->383 384 # of days in month <= 30.5 mae = 4783.12 samples = 10 value = 35412.0 383->384 387 # of 0 bedrooms * days in month <= 29.0 mae = 3540.418 samples = 17 value = 41198.299 383->387 385 mae = 5053.9 samples = 4 value = 33329.2 384->385 386 mae = 4111.9 samples = 6 value = 37428.75 384->386 388 # of 1 bedrooms * days in month <= 549.0 mae = 2771.546 samples = 13 value = 40274.401 387->388 391 mae = 4212.025 samples = 4 value = 43276.0 387->391 389 mae = 2438.1 samples = 5 value = 38554.2 388->389 390 mae = 2764.925 samples = 8 value = 40585.65 388->390 393 1 bedrooms days vacant <= 15.5 mae = 2685.231 samples = 13 value = 36123.5 392->393 396 mae = 4274.8 samples = 8 value = 38392.3 392->396 394 mae = 2406.3 samples = 9 value = 36123.5 393->394 395 mae = 1865.125 samples = 4 value = 32760.1 393->395 399 1 bedrooms days vacant <= 27.5 mae = 7561.549 samples = 62 value = 33960.4 398->399 418 mae = 5647.066 samples = 6 value = 38158.1 398->418 400 bldg_age <= 126.5 mae = 8231.07 samples = 55 value = 34469.501 399->400 417 mae = 1349.157 samples = 7 value = 32576.2 399->417 401 0 bedrooms - days vacant <= 8.0 mae = 7115.591 samples = 48 value = 34675.55 400->401 416 mae = 14913.214 samples = 7 value = 29381.501 400->416 402 month_september <= 0.5 mae = 6512.643 samples = 36 value = 34780.3 401->402 413 bldg_age <= 124.5 mae = 8653.133 samples = 12 value = 33669.15 401->413 403 bldg_age <= 124.5 mae = 3852.695 samples = 31 value = 34765.4 402->403 412 mae = 22150.62 samples = 5 value = 39033.9 402->412 404 bldg_age <= 100.5 mae = 2579.8 samples = 24 value = 35034.45 403->404 411 mae = 7680.351 samples = 7 value = 31009.501 403->411 405 mae = 3627.02 samples = 5 value = 36051.3 404->405 406 if multi-family, # units <= 17.0 mae = 2236.537 samples = 19 value = 34765.4 404->406 407 bldg_age <= 123.5 mae = 1989.192 samples = 13 value = 35303.501 406->407 410 mae = 2185.1 samples = 6 value = 33179.4 406->410 408 mae = 1559.0 samples = 7 value = 35303.501 407->408 409 mae = 2491.083 samples = 6 value = 35357.0 407->409 414 mae = 1473.24 samples = 5 value = 30853.6 413->414 415 mae = 12879.257 samples = 7 value = 34114.0 413->415 420 # of 2 bedrooms * days in month <= 152.5 mae = 8814.625 samples = 81 value = 36292.9 419->420 445 2 bedrooms days vacant <= 0.5 mae = 3989.008 samples = 97 value = 27996.5 419->445 421 bldg_age <= 50.5 mae = 8158.639 samples = 61 value = 34294.799 420->421 440 2 bedrooms days vacant <= 0.5 mae = 8107.295 samples = 20 value = 44340.1 420->440 422 2 bedrooms days vacant <= 0.5 mae = 8971.863 samples = 8 value = 47272.0 421->422 425 bldg_age <= 51.5 mae = 7129.328 samples = 53 value = 32679.5 421->425 423 mae = 9038.35 samples = 4 value = 38267.75 422->423 424 mae = 4510.025 samples = 4 value = 51007.6 422->424 426 # of 5 bedrooms * days in month <= 30.5 mae = 5219.47 samples = 10 value = 28365.65 425->426 429 # of 2 bedrooms * days in month <= 122.0 mae = 7082.619 samples = 43 value = 34294.799 425->429 427 mae = 2301.925 samples = 4 value = 27349.8 426->427 428 mae = 6786.467 samples = 6 value = 29292.25 426->428 430 mae = 2951.4 samples = 9 value = 31158.4 429->430 431 2 bedrooms days vacant <= 0.5 mae = 7720.271 samples = 34 value = 35749.9 429->431 432 bldg_age <= 73.5 mae = 7021.224 samples = 29 value = 37947.3 431->432 439 mae = 8649.22 samples = 5 value = 30621.7 431->439 433 # of 5 bedrooms * days in month <= 15.5 mae = 6638.752 samples = 21 value = 38217.2 432->433 438 mae = 6674.525 samples = 8 value = 31846.05 432->438 434 mae = 4404.912 samples = 8 value = 42356.75 433->434 435 bldg_age <= 52.5 mae = 6843.077 samples = 13 value = 35206.9 433->435 436 mae = 11736.3 samples = 5 value = 39284.5 435->436 437 mae = 2451.213 samples = 8 value = 34473.25 435->437 441 month_march <= 0.5 mae = 5250.169 samples = 16 value = 46642.4 440->441 444 mae = 2924.7 samples = 4 value = 24273.4 440->444 442 mae = 5520.808 samples = 12 value = 45182.1 441->442 443 mae = 3735.75 samples = 4 value = 48101.3 441->443 446 month_april <= 0.5 mae = 3779.01 samples = 89 value = 28329.5 445->446 479 mae = 4619.813 samples = 8 value = 20998.4 445->479 447 bldg_age <= 110.5 mae = 3607.067 samples = 81 value = 27740.7 446->447 476 bldg_age <= 76.5 mae = 3905.613 samples = 8 value = 31519.7 446->476 448 # of 2 bedrooms * days in month <= 142.5 mae = 3677.651 samples = 47 value = 26032.4 447->448 465 month_august <= 0.5 mae = 3083.091 samples = 34 value = 29145.05 447->465 449 # of days in month <= 30.5 mae = 2821.5 samples = 10 value = 23969.05 448->449 452 month_november <= 0.5 mae = 3500.516 samples = 37 value = 28329.5 448->452 450 mae = 2698.66 samples = 5 value = 24149.7 449->450 451 mae = 2872.08 samples = 5 value = 23788.4 449->451 453 month_october <= 0.5 mae = 3389.864 samples = 33 value = 28460.6 452->453 464 mae = 2718.125 samples = 4 value = 24751.15 452->464 454 month_may <= 0.5 mae = 3226.052 samples = 29 value = 28512.9 453->454 463 mae = 2925.675 samples = 4 value = 24953.85 453->463 455 bldg_age <= 75.5 mae = 3211.908 samples = 24 value = 28035.1 454->455 462 mae = 2549.16 samples = 5 value = 31524.2 454->462 456 mae = 2499.456 samples = 9 value = 29511.8 455->456 457 # of 2 bedrooms * days in month <= 152.5 mae = 3310.76 samples = 15 value = 26070.5 455->457 458 mae = 3345.717 samples = 6 value = 27578.8 457->458 459 bldg_age <= 76.5 mae = 3240.867 samples = 9 value = 25651.2 457->459 460 mae = 2722.72 samples = 5 value = 25651.2 459->460 461 mae = 3888.55 samples = 4 value = 26285.75 459->461 466 month_march <= 0.5 mae = 2930.66 samples = 30 value = 28633.5 465->466 475 mae = 1369.425 samples = 4 value = 32974.35 465->475 467 bldg_age <= 111.5 mae = 2958.585 samples = 26 value = 28295.7 466->467 474 mae = 2175.2 samples = 4 value = 30839.2 466->474 468 mae = 4278.171 samples = 7 value = 31171.9 467->468 469 # of 3 bedrooms * days in month <= 122.0 mae = 2283.453 samples = 19 value = 27581.5 467->469 470 mae = 1123.038 samples = 8 value = 27349.25 469->470 471 bldg_age <= 112.5 mae = 2806.918 samples = 11 value = 28672.1 469->471 472 mae = 3588.125 samples = 4 value = 25602.9 471->472 473 mae = 1882.071 samples = 7 value = 29550.4 471->473 477 mae = 2979.025 samples = 4 value = 31519.7 476->477 478 mae = 4832.2 samples = 4 value = 29955.75 476->478 481 bldg_age <= 115.5 mae = 5005.107 samples = 91 value = 22113.3 480->481 512 bldg_age <= 106.5 mae = 2913.894 samples = 35 value = 26206.1 480->512 482 bldg_age <= 13.5 mae = 4773.313 samples = 78 value = 21679.05 481->482 509 # of 3 bedrooms * days in month <= 122.0 mae = 4363.446 samples = 13 value = 27422.2 481->509 483 bldg_age <= 11.5 mae = 5614.316 samples = 38 value = 24403.7 482->483 496 bldg_age <= 113.5 mae = 3138.75 samples = 40 value = 19897.8 482->496 484 bldg_age <= 10.5 mae = 5952.216 samples = 19 value = 22528.5 483->484 489 bldg_age <= 12.5 mae = 4614.463 samples = 19 value = 26698.4 483->489 485 mae = 5084.7 samples = 9 value = 24975.6 484->485 486 # of days in month <= 30.5 mae = 6295.31 samples = 10 value = 19699.45 484->486 487 mae = 4825.72 samples = 5 value = 22442.0 486->487 488 mae = 6667.88 samples = 5 value = 16956.9 486->488 490 # of 3 bedrooms * days in month <= 91.5 mae = 5340.3 samples = 10 value = 27497.05 489->490 493 # of 3 bedrooms * days in month <= 91.5 mae = 2516.233 samples = 9 value = 22909.7 489->493 491 mae = 7491.58 samples = 5 value = 26712.8 490->491 492 mae = 2952.54 samples = 5 value = 27895.2 490->492 494 mae = 2038.125 samples = 4 value = 23071.45 493->494 495 mae = 2898.72 samples = 5 value = 22909.7 493->495 497 # stories <= 2.25 mae = 2737.0 samples = 22 value = 18536.95 496->497 504 bldg_age <= 114.5 mae = 3213.133 samples = 18 value = 20800.7 496->504 498 heating system_furnace (high efficiency condensing)  <= 0.5 mae = 2651.717 samples = 12 value = 20883.85 497->498 501 # of 3 bedrooms * days in month <= 122.0 mae = 1429.48 samples = 10 value = 17012.3 497->501 499 mae = 3896.74 samples = 5 value = 21263.4 498->499 500 mae = 1535.7 samples = 7 value = 19676.4 498->500 502 mae = 933.975 samples = 4 value = 17012.3 501->502 503 mae = 1759.817 samples = 6 value = 16707.9 501->503 505 # of days in month <= 30.5 mae = 1605.867 samples = 9 value = 20545.0 504->505 508 mae = 4646.145 samples = 9 value = 22113.3 504->508 506 mae = 1487.6 samples = 4 value = 21174.7 505->506 507 mae = 1531.48 samples = 5 value = 19700.0 505->507 510 mae = 4610.375 samples = 4 value = 24104.1 509->510 511 mae = 3659.567 samples = 9 value = 29729.0 509->511 513 # of 3 bedrooms * days in month <= 30.5 mae = 2797.35 samples = 14 value = 27769.7 512->513 516 bldg_age <= 107.5 mae = 2579.567 samples = 21 value = 24657.0 512->516 514 mae = 1616.7 samples = 6 value = 28913.4 513->514 515 mae = 3277.462 samples = 8 value = 26669.4 513->515 517 mae = 2857.588 samples = 8 value = 22217.35 516->517 518 bldg_age <= 108.5 mae = 2090.277 samples = 13 value = 26061.4 516->518 519 mae = 1556.771 samples = 7 value = 25023.8 518->519 520 mae = 2475.433 samples = 6 value = 26302.65 518->520 522 basement sq ft <= 6926.5 mae = 13623.593 samples = 315 value = 92493.2 521->522 645 bldg_age <= 46.5 mae = 20370.539 samples = 437 value = 74399.998 521->645 523 bldg_age <= 95.5 mae = 12413.766 samples = 268 value = 89942.2 522->523 630 bldg_age <= 97.5 mae = 11388.481 samples = 47 value = 109016.001 522->630 524 # of days in month <= 29.5 mae = 11860.66 samples = 156 value = 93604.451 523->524 585 bldg_age <= 98.5 mae = 11351.429 samples = 112 value = 84475.6 523->585 525 bldg_age <= 51.5 mae = 8436.323 samples = 13 value = 82130.301 524->525 528 month_august <= 0.5 mae = 11897.664 samples = 143 value = 94008.201 524->528 526 mae = 6087.371 samples = 7 value = 88293.001 525->526 527 mae = 8590.217 samples = 6 value = 78769.55 525->527 529 month_january <= 0.5 mae = 11439.069 samples = 130 value = 93604.451 528->529 580 basement sq ft <= 5365.5 mae = 13882.046 samples = 13 value = 101758.999 528->580 530 2 bedrooms days vacant <= 14.5 mae = 11548.667 samples = 115 value = 94232.802 529->530 575 1 bedrooms days vacant <= 2.5 mae = 8675.733 samples = 15 value = 89114.5 529->575 531 bldg_age <= 53.5 mae = 11264.725 samples = 96 value = 94901.25 530->531 568 bldg_age <= 51.5 mae = 10993.958 samples = 19 value = 84380.3 530->568 532 3 bedrooms days vacant <= 29.0 mae = 11047.234 samples = 67 value = 93254.999 531->532 559 bldg_age <= 74.5 mae = 10460.189 samples = 29 value = 101580.001 531->559 533 month_april <= 0.5 mae = 10873.627 samples = 62 value = 92551.8 532->533 558 mae = 7283.359 samples = 5 value = 101222.0 532->558 534 month_december <= 0.5 mae = 10958.082 samples = 56 value = 92972.349 533->534 557 mae = 6993.083 samples = 6 value = 85836.3 533->557 535 month_november <= 0.5 mae = 10211.306 samples = 46 value = 93802.101 534->535 554 bldg_age <= 50.5 mae = 12957.55 samples = 10 value = 87565.7 534->554 536 1 bedrooms days vacant <= 0.5 mae = 9248.739 samples = 38 value = 94191.401 535->536 551 total # bedrooms  in building <= 31.5 mae = 12913.6 samples = 8 value = 87999.051 535->551 537 2 bedrooms days vacant <= 0.5 mae = 8888.761 samples = 23 value = 99080.901 536->537 546 bldg_age <= 50.5 mae = 8546.774 samples = 15 value = 91934.701 536->546 538 bldg_age <= 8.5 mae = 9528.052 samples = 19 value = 100480.001 537->538 545 mae = 2425.351 samples = 4 value = 93747.5 537->545 539 mae = 3647.701 samples = 6 value = 93092.25 538->539 540 month_september <= 0.5 mae = 10667.523 samples = 13 value = 102741.001 538->540 541 bldg_age <= 52.5 mae = 9651.866 samples = 9 value = 102699.999 540->541 544 mae = 12480.5 samples = 4 value = 110398.0 540->544 542 mae = 6721.16 samples = 5 value = 95991.801 541->542 543 mae = 11638.199 samples = 4 value = 110180.499 541->543 547 1 bedrooms days vacant <= 29.0 mae = 6480.77 samples = 10 value = 92602.401 546->547 550 mae = 10987.16 samples = 5 value = 88772.001 546->550 548 mae = 6905.374 samples = 4 value = 98074.901 547->548 549 mae = 5625.134 samples = 6 value = 91804.551 547->549 552 mae = 20769.275 samples = 4 value = 86212.0 551->552 553 mae = 5057.925 samples = 4 value = 87999.051 551->553 555 mae = 18087.08 samples = 5 value = 89901.201 554->555 556 mae = 6919.8 samples = 5 value = 85360.099 554->556 560 # of 2 bedrooms * days in month <= 213.5 mae = 7221.186 samples = 14 value = 104486.0 559->560 563 0 bedrooms - days vacant <= 37.0 mae = 11272.547 samples = 15 value = 94848.1 559->563 561 mae = 6859.979 samples = 5 value = 107633.001 560->561 562 mae = 7053.966 samples = 9 value = 104321.999 560->562 564 0 bedrooms - days vacant <= 5.5 mae = 9938.736 samples = 11 value = 95374.1 563->564 567 mae = 12962.675 samples = 4 value = 87543.45 563->567 565 mae = 1496.05 samples = 4 value = 95111.1 564->565 566 mae = 13832.286 samples = 7 value = 101890.0 564->566 569 # of 3 bedrooms * days in month <= 152.5 mae = 13469.491 samples = 11 value = 95070.202 568->569 572 # of 2 bedrooms * days in month <= 288.5 mae = 4443.162 samples = 8 value = 83549.151 568->572 570 mae = 10880.26 samples = 5 value = 102006.001 569->570 571 mae = 12569.451 samples = 6 value = 84823.801 569->571 573 mae = 7461.175 samples = 4 value = 81923.899 572->573 574 mae = 1425.15 samples = 4 value = 83549.151 572->574 576 bldg_age <= 51.5 mae = 7932.778 samples = 9 value = 84698.899 575->576 579 mae = 7395.733 samples = 6 value = 91559.55 575->579 577 mae = 9954.55 samples = 4 value = 89184.65 576->577 578 mae = 4309.76 samples = 5 value = 81079.499 576->578 581 bldg_age <= 51.5 mae = 16185.566 samples = 9 value = 101494.001 580->581 584 mae = 7993.375 samples = 4 value = 105210.0 580->584 582 mae = 20624.024 samples = 4 value = 111747.0 581->582 583 mae = 12099.0 samples = 5 value = 98815.0 581->583 586 # of 2 bedrooms * days in month <= 290.0 mae = 9886.606 samples = 63 value = 86315.5 585->586 611 2 bedrooms days vacant <= 38.0 mae = 11506.579 samples = 49 value = 79307.799 585->611 587 1 bedrooms days vacant <= 2.0 mae = 9543.439 samples = 46 value = 85078.8 586->587 606 2 bedrooms days vacant <= 4.0 mae = 6818.894 samples = 17 value = 94705.699 586->606 588 month_october <= 0.5 mae = 7556.389 samples = 36 value = 85977.95 587->588 603 # of days in month <= 30.5 mae = 10867.9 samples = 10 value = 74433.6 587->603 589 month_april <= 0.5 mae = 7395.666 samples = 32 value = 86291.5 588->589 602 mae = 3892.575 samples = 4 value = 76069.599 588->602 590 month_march <= 0.5 mae = 7700.768 samples = 28 value = 86581.6 589->590 601 mae = 2229.25 samples = 4 value = 80322.2 589->601 591 bldg_age <= 96.5 mae = 5620.883 samples = 24 value = 86882.6 590->591 600 mae = 17428.376 samples = 4 value = 75591.0 590->600 592 mae = 5213.6 samples = 7 value = 92020.3 591->592 593 bldg_age <= 97.5 mae = 5143.117 samples = 17 value = 86315.5 591->593 594 # of 0 bedrooms * days in month <= 531.0 mae = 2211.189 samples = 9 value = 86016.7 593->594 597 # of days in month <= 30.5 mae = 8259.137 samples = 8 value = 86882.6 593->597 595 mae = 2560.62 samples = 5 value = 84344.801 594->595 596 mae = 525.875 samples = 4 value = 86291.5 594->596 598 mae = 10584.525 samples = 4 value = 88317.5 597->598 599 mae = 5933.749 samples = 4 value = 86258.35 597->599 604 mae = 4449.36 samples = 5 value = 66481.299 603->604 605 mae = 15274.46 samples = 5 value = 76541.2 603->605 607 # of days in month <= 30.5 mae = 7716.236 samples = 11 value = 95900.6 606->607 610 mae = 3917.883 samples = 6 value = 90389.55 606->610 608 mae = 10753.54 samples = 5 value = 94879.101 607->608 609 mae = 4361.0 samples = 6 value = 97988.95 607->609 612 1 bedrooms days vacant <= 23.5 mae = 10801.755 samples = 45 value = 79888.999 611->612 629 mae = 3833.05 samples = 4 value = 59254.95 611->629 613 bldg_age <= 99.5 mae = 6700.771 samples = 35 value = 78032.5 612->613 626 bldg_age <= 100.5 mae = 19697.25 samples = 10 value = 101002.65 612->626 614 2 bedrooms days vacant <= 4.0 mae = 5233.343 samples = 14 value = 73694.351 613->614 619 # of 2 bedrooms * days in month <= 305.0 mae = 6947.29 samples = 21 value = 79952.2 613->619 615 # of 2 bedrooms * days in month <= 290.0 mae = 5587.17 samples = 10 value = 71796.2 614->615 618 mae = 2530.124 samples = 4 value = 76432.3 614->618 616 mae = 6039.32 samples = 5 value = 77879.701 615->616 617 mae = 3643.7 samples = 5 value = 70423.101 615->617 620 1 bedrooms days vacant <= 0.5 mae = 6088.643 samples = 14 value = 79927.099 619->620 625 mae = 8335.6 samples = 7 value = 82255.1 619->625 621 if multi-family sum of apt sq ft <= 14227.5 mae = 4003.333 samples = 9 value = 79952.2 620->621 624 mae = 9408.1 samples = 5 value = 77781.701 620->624 622 mae = 1542.22 samples = 5 value = 79901.999 621->622 623 mae = 6543.725 samples = 4 value = 87083.8 621->623 627 mae = 8498.259 samples = 5 value = 83911.8 626->627 628 mae = 7842.2 samples = 5 value = 125644.0 626->628 631 # of 3 bedrooms * days in month <= 61.0 mae = 14605.0 samples = 11 value = 98057.8 630->631 634 bldg_age <= 98.5 mae = 9164.739 samples = 36 value = 110125.999 630->634 632 mae = 13158.14 samples = 5 value = 93707.799 631->632 633 mae = 15085.717 samples = 6 value = 100540.901 631->633 635 # of 3 bedrooms * days in month <= 61.0 mae = 10313.283 samples = 12 value = 122149.5 634->635 638 bldg_age <= 99.5 mae = 6608.217 samples = 24 value = 109301.5 634->638 636 mae = 13815.08 samples = 5 value = 122701.001 635->636 637 mae = 7654.428 samples = 7 value = 121598.0 635->637 639 mae = 9634.563 samples = 8 value = 105792.001 638->639 640 bldg_age <= 100.5 mae = 4713.794 samples = 16 value = 109618.499 638->640 641 # of days in month <= 30.5 mae = 4957.973 samples = 11 value = 108976.0 640->641 644 mae = 3268.6 samples = 5 value = 113390.001 640->644 642 mae = 1765.0 samples = 4 value = 109195.499 641->642 643 mae = 6782.529 samples = 7 value = 108976.0 641->643 646 basement sq ft <= 3419.5 mae = 17874.419 samples = 236 value = 81544.4 645->646 737 bldg_age <= 99.5 mae = 19061.985 samples = 201 value = 60810.0 645->737 647 bldg_age <= 28.5 mae = 13670.003 samples = 149 value = 76630.401 646->647 704 bldg_age <= 42.5 mae = 21757.04 samples = 87 value = 91655.398 646->704 648 bldg_age <= 24.5 mae = 13585.646 samples = 67 value = 84970.801 647->648 671 # of 1 bedrooms * days in month <= 754.0 mae = 11291.546 samples = 82 value = 70375.849 647->671 649 1 bedrooms days vacant <= 169.5 mae = 9293.955 samples = 40 value = 75593.75 648->649 662 0 bedrooms - days vacant <= 54.5 mae = 12083.03 samples = 27 value = 94090.898 648->662 650 1 bedrooms days vacant <= 29.0 mae = 8869.014 samples = 34 value = 76706.249 649->650 661 mae = 7026.417 samples = 6 value = 62333.302 649->661 651 mae = 5069.98 samples = 5 value = 70471.898 650->651 652 bldg_age <= 21.5 mae = 8767.193 samples = 29 value = 79621.1 650->652 653 mae = 5185.224 samples = 8 value = 73917.999 652->653 654 bldg_age <= 22.5 mae = 9026.986 samples = 21 value = 80877.798 652->654 655 mae = 10870.385 samples = 7 value = 92347.399 654->655 656 1 bedrooms days vacant <= 121.0 mae = 4368.043 samples = 14 value = 78821.799 654->656 657 1 bedrooms days vacant <= 96.5 mae = 3098.763 samples = 8 value = 77537.549 656->657 660 mae = 5307.082 samples = 6 value = 80580.15 656->660 658 mae = 2832.175 samples = 4 value = 77537.549 657->658 659 mae = 3365.351 samples = 4 value = 77224.35 657->659 663 0 bedrooms - days vacant <= 14.0 mae = 7423.341 samples = 12 value = 93211.448 662->663 666 bldg_age <= 27.5 mae = 15398.72 samples = 15 value = 97509.201 662->666 664 mae = 8365.543 samples = 7 value = 94090.898 663->664 665 mae = 3429.521 samples = 5 value = 88717.202 663->665 667 # of 0 bedrooms * days in month <= 1159.0 mae = 14637.112 samples = 8 value = 116285.5 666->667 670 mae = 7938.843 samples = 7 value = 93912.201 666->670 668 mae = 14321.2 samples = 4 value = 119847.501 667->668 669 mae = 14953.025 samples = 4 value = 112000.001 667->669 672 0 bedrooms - days vacant <= 30.0 mae = 7368.49 samples = 32 value = 64985.599 671->672 683 1 bedrooms days vacant <= 58.5 mae = 11518.194 samples = 50 value = 76956.5 671->683 673 bldg_age <= 29.5 mae = 8250.786 samples = 22 value = 64075.049 672->673 680 bldg_age <= 29.5 mae = 4741.599 samples = 10 value = 66685.351 672->680 674 mae = 11676.5 samples = 4 value = 58067.5 673->674 675 if multi-family, # units <= 32.0 mae = 6906.628 samples = 18 value = 64346.35 673->675 676 mae = 4341.58 samples = 5 value = 65752.799 675->676 677 0 bedrooms - days vacant <= 3.5 mae = 7460.169 samples = 13 value = 63688.6 675->677 678 mae = 7425.185 samples = 7 value = 56454.601 677->678 679 mae = 5391.817 samples = 6 value = 64241.599 677->679 681 mae = 3276.899 samples = 6 value = 67414.1 680->681 682 mae = 6314.249 samples = 4 value = 61312.449 680->682 684 month_march <= 0.5 mae = 11081.213 samples = 46 value = 77391.3 683->684 703 mae = 7554.325 samples = 4 value = 58347.8 683->703 685 # of 1 bedrooms * days in month <= 793.0 mae = 11617.214 samples = 42 value = 78396.75 684->685 702 mae = 2089.9 samples = 4 value = 71927.001 684->702 686 1 bedrooms days vacant <= 0.5 mae = 11647.911 samples = 18 value = 74888.25 685->686 693 bldg_age <= 45.5 mae = 10979.15 samples = 24 value = 80027.151 685->693 687 mae = 9260.859 samples = 5 value = 70449.4 686->687 688 1 bedrooms days vacant <= 12.5 mae = 11237.7 samples = 13 value = 77282.6 686->688 689 mae = 11794.875 samples = 4 value = 84782.6 688->689 690 1 bedrooms days vacant <= 26.0 mae = 10337.889 samples = 9 value = 76630.401 688->690 691 mae = 9756.05 samples = 4 value = 61903.4 690->691 692 mae = 6717.4 samples = 5 value = 77282.6 690->692 694 1 bedrooms days vacant <= 0.5 mae = 7457.98 samples = 15 value = 79184.799 693->694 699 1 bedrooms days vacant <= 25.5 mae = 14376.744 samples = 9 value = 91315.201 693->699 695 mae = 4650.0 samples = 5 value = 77499.999 694->695 696 1 bedrooms days vacant <= 20.0 mae = 8693.491 samples = 10 value = 79521.75 694->696 697 mae = 6289.85 samples = 6 value = 79858.701 696->697 698 mae = 11962.0 samples = 4 value = 76489.149 696->698 700 mae = 8339.599 samples = 4 value = 91989.1 699->700 701 mae = 17454.3 samples = 5 value = 82554.399 699->701 705 0 bedrooms - days vacant <= 25.0 mae = 23096.475 samples = 61 value = 95437.202 704->705 728 1 bedrooms days vacant <= 62.5 mae = 14583.158 samples = 26 value = 81885.8 704->728 706 1 bedrooms days vacant <= 17.5 mae = 22313.053 samples = 19 value = 138905.999 705->706 711 # of 0 bedrooms * days in month <= 1586.0 mae = 10911.833 samples = 42 value = 89613.651 705->711 707 mae = 15228.0 samples = 7 value = 125673.002 706->707 708 1 bedrooms days vacant <= 37.5 mae = 20534.084 samples = 12 value = 151152.0 706->708 709 mae = 12967.801 samples = 5 value = 148887.001 708->709 710 mae = 25291.429 samples = 7 value = 153416.999 708->710 712 0 bedrooms - days vacant <= 94.0 mae = 7742.634 samples = 18 value = 85667.951 711->712 719 0 bedrooms - days vacant <= 178.0 mae = 11927.167 samples = 24 value = 92693.65 711->719 713 mae = 7200.2 samples = 5 value = 93331.3 712->713 714 bldg_age <= 9.5 mae = 6227.378 samples = 13 value = 80783.399 712->714 715 mae = 2655.8 samples = 4 value = 76786.549 714->715 716 0 bedrooms - days vacant <= 124.0 mae = 5399.0 samples = 9 value = 85744.5 714->716 717 mae = 3743.401 samples = 4 value = 82844.85 716->717 718 mae = 6367.82 samples = 5 value = 87216.6 716->718 720 bldg_age <= 8.5 mae = 9205.906 samples = 18 value = 95078.3 719->720 727 mae = 17868.316 samples = 6 value = 83974.751 719->727 721 mae = 6093.8 samples = 5 value = 91929.9 720->721 722 0 bedrooms - days vacant <= 117.0 mae = 9470.262 samples = 13 value = 96828.602 720->722 723 mae = 11967.759 samples = 5 value = 108605.999 722->723 724 bldg_age <= 10.5 mae = 5640.851 samples = 8 value = 95750.499 722->724 725 mae = 1857.5 samples = 4 value = 97687.301 724->725 726 mae = 6837.6 samples = 4 value = 87835.998 724->726 729 bldg_age <= 43.5 mae = 13311.747 samples = 19 value = 86175.499 728->729 736 mae = 6182.972 samples = 7 value = 65457.0 728->736 730 mae = 24443.499 samples = 6 value = 113489.0 729->730 731 2 bedrooms days vacant <= 8.5 mae = 5088.308 samples = 13 value = 82692.301 729->731 732 1 bedrooms days vacant <= 30.5 mae = 4458.601 samples = 9 value = 81079.299 731->732 735 mae = 4222.749 samples = 4 value = 87210.549 731->735 733 mae = 5347.376 samples = 4 value = 83171.449 732->733 734 mae = 3236.02 samples = 5 value = 78521.499 732->734 738 1 bedrooms days vacant <= 29.5 mae = 19282.228 samples = 175 value = 64348.899 737->738 805 0 bedrooms - days vacant <= 9.0 mae = 5194.789 samples = 26 value = 43764.2 737->805 739 0 bedrooms - days vacant <= 40.5 mae = 17146.433 samples = 142 value = 61657.349 738->739 794 heating system_boiler (hot water) <= 0.5 mae = 24541.612 samples = 33 value = 75707.301 738->794 740 if multi-family sum of apt sq ft <= 13007.0 mae = 18625.577 samples = 110 value = 63510.501 739->740 781 0 bedrooms - days vacant <= 94.0 mae = 10792.075 samples = 32 value = 57101.35 739->781 741 0 bedrooms - days vacant <= 17.0 mae = 15010.4 samples = 27 value = 73047.099 740->741 752 # of 0 bedrooms * days in month <= 725.0 mae = 19103.713 samples = 83 value = 61662.999 740->752 742 # of 1 bedrooms * days in month <= 122.0 mae = 10774.918 samples = 17 value = 63412.101 741->742 749 bldg_age <= 95.5 mae = 17229.98 samples = 10 value = 79755.3 741->749 743 mae = 5449.9 samples = 4 value = 54788.8 742->743 744 bldg_age <= 98.5 mae = 9897.177 samples = 13 value = 71330.4 742->744 745 gross bldg sq ft <= 18528.0 mae = 7938.411 samples = 9 value = 73047.099 744->745 748 mae = 7626.025 samples = 4 value = 57922.4 744->748 746 mae = 5202.4 samples = 4 value = 76398.0 745->746 747 mae = 9783.88 samples = 5 value = 71330.4 745->747 750 mae = 5691.466 samples = 6 value = 77779.401 749->750 751 mae = 14995.1 samples = 4 value = 120942.5 749->751 753 1 bedrooms days vacant <= 0.5 mae = 21495.978 samples = 54 value = 59668.4 752->753 774 bldg_age <= 96.5 mae = 13514.973 samples = 29 value = 65832.801 752->774 754 bldg_age <= 47.5 mae = 18309.986 samples = 21 value = 46876.401 753->754 761 bldg_age <= 47.5 mae = 22127.924 samples = 33 value = 62934.799 753->761 755 # of days in month <= 30.5 mae = 18912.04 samples = 10 value = 38614.15 754->755 758 # of 1 bedrooms * days in month <= 767.0 mae = 10288.464 samples = 11 value = 58630.4 754->758 756 mae = 27386.017 samples = 6 value = 37989.15 755->756 757 mae = 6065.225 samples = 4 value = 39255.45 755->757 759 mae = 5409.551 samples = 4 value = 60754.85 758->759 760 mae = 12294.0 samples = 7 value = 56739.101 758->760 762 1 bedrooms days vacant <= 3.0 mae = 29145.677 samples = 13 value = 82217.4 761->762 767 # of days in month <= 29.5 mae = 11601.215 samples = 20 value = 60662.8 761->767 763 mae = 30335.42 samples = 5 value = 83228.301 762->763 764 1 bedrooms days vacant <= 12.0 mae = 27786.6 samples = 8 value = 79754.9 762->764 765 mae = 41864.175 samples = 4 value = 78010.899 764->765 766 mae = 13709.025 samples = 4 value = 80591.85 764->766 768 mae = 4958.251 samples = 4 value = 49826.15 767->768 769 1 bedrooms days vacant <= 12.5 mae = 11051.344 samples = 16 value = 61662.999 767->769 770 1 bedrooms days vacant <= 3.5 mae = 12372.209 samples = 11 value = 65032.599 769->770 773 mae = 6275.86 samples = 5 value = 59662.899 769->773 771 mae = 10147.657 samples = 7 value = 61662.999 770->771 772 mae = 10705.375 samples = 4 value = 73807.35 770->772 775 mae = 15011.938 samples = 13 value = 78238.199 774->775 776 bldg_age <= 98.5 mae = 7452.375 samples = 16 value = 59591.5 774->776 777 0 bedrooms - days vacant <= 7.0 mae = 6110.909 samples = 12 value = 61997.1 776->777 780 mae = 9480.175 samples = 4 value = 48895.85 776->780 778 mae = 6026.5 samples = 6 value = 59549.2 777->778 779 mae = 4015.15 samples = 6 value = 64272.2 777->779 782 # of days in month <= 30.5 mae = 10005.557 samples = 23 value = 55138.101 781->782 791 0 bedrooms - days vacant <= 141.0 mae = 9438.889 samples = 9 value = 64501.101 781->791 783 0 bedrooms - days vacant <= 59.0 mae = 12007.829 samples = 10 value = 58513.5 782->783 786 0 bedrooms - days vacant <= 82.0 mae = 7830.047 samples = 13 value = 53559.0 782->786 784 mae = 9901.775 samples = 4 value = 54469.75 783->784 785 mae = 12073.8 samples = 6 value = 60411.3 783->785 787 bldg_age <= 97.5 mae = 6108.023 samples = 9 value = 47370.999 786->787 790 mae = 8969.551 samples = 4 value = 56171.15 786->790 788 mae = 5453.46 samples = 5 value = 53559.0 787->788 789 mae = 5379.225 samples = 4 value = 45769.65 787->789 792 mae = 13917.58 samples = 5 value = 67110.0 791->792 793 mae = 3188.3 samples = 4 value = 64425.0 791->793 795 bldg_age <= 96.0 mae = 27186.673 samples = 26 value = 79418.099 794->795 804 mae = 7055.672 samples = 7 value = 60857.701 794->804 796 # of 1 bedrooms * days in month <= 793.0 mae = 26950.677 samples = 22 value = 75761.25 795->796 803 mae = 13899.05 samples = 4 value = 110792.5 795->803 797 bldg_age <= 48.5 mae = 22291.309 samples = 12 value = 72932.0 796->797 800 1 bedrooms days vacant <= 33.0 mae = 31315.56 samples = 10 value = 82158.7 796->800 798 mae = 35056.52 samples = 5 value = 73043.5 797->798 799 mae = 13141.443 samples = 7 value = 72820.5 797->799 801 mae = 40704.3 samples = 5 value = 85250.001 800->801 802 mae = 20690.3 samples = 5 value = 79067.399 800->802 806 1 bedrooms days vacant <= 25.5 mae = 6812.389 samples = 9 value = 47850.899 805->806 809 1 bedrooms days vacant <= 28.5 mae = 3373.894 samples = 17 value = 41563.501 805->809 807 mae = 5698.25 samples = 4 value = 46940.65 806->807 808 mae = 7567.94 samples = 5 value = 48529.699 806->808 810 mae = 3835.572 samples = 7 value = 45502.301 809->810 811 1 bedrooms days vacant <= 41.5 mae = 1775.38 samples = 10 value = 40078.8 809->811 812 mae = 1907.32 samples = 5 value = 41563.501 811->812 813 mae = 1313.86 samples = 5 value = 39915.6 811->813 815 total # bedrooms  in building <= 76.0 mae = 55644.755 samples = 958 value = 172613.0 814->815 1192 common laundry facilities_no <= 0.5 mae = 58082.125 samples = 145 value = 404376.005 814->1192 816 if multi-family sum of apt sq ft <= 30084.0 mae = 39981.277 samples = 464 value = 136590.528 815->816 997 if environmental certification: type_other (type in) <= 0.5 mae = 52692.777 samples = 494 value = 204993.998 815->997 817 if environmental certification: type_leed, platinum <= 0.5 mae = 35614.382 samples = 328 value = 117267.5 816->817 944 bldg_age <= 9.5 mae = 31538.349 samples = 136 value = 173733.289 816->944 818 if multi-family sum of apt sq ft <= 29767.0 mae = 35208.842 samples = 287 value = 125019.0 817->818 927 0 bedrooms - days vacant <= 110.0 mae = 6514.636 samples = 41 value = 82842.88 817->927 819 bldg_age <= 114.5 mae = 33109.299 samples = 253 value = 130016.0 818->819 914 1 bedrooms days vacant <= 72.5 mae = 19611.665 samples = 34 value = 71566.6 818->914 820 basement sq ft <= 6515.0 mae = 33899.145 samples = 210 value = 137934.499 819->820 897 if multi-family sum of apt sq ft <= 26657.5 mae = 13480.105 samples = 43 value = 107502.002 819->897 821 bldg_age <= 27.5 mae = 25739.508 samples = 120 value = 128795.0 820->821 866 0 bedrooms - days vacant <= 83.0 mae = 40127.99 samples = 90 value = 151229.003 820->866 822 0 bedrooms - days vacant <= 147.5 mae = 27360.434 samples = 53 value = 138033.999 821->822 841 # of 0 bedrooms * days in month <= 1820.0 mae = 21491.861 samples = 67 value = 120773.828 821->841 823 month_december <= 0.5 mae = 23173.562 samples = 48 value = 137156.499 822->823 840 mae = 25143.601 samples = 5 value = 210640.0 822->840 824 # of 2 bedrooms * days in month <= 280.0 mae = 19768.818 samples = 44 value = 134393.501 823->824 839 mae = 22724.25 samples = 4 value = 183649.0 823->839 825 0 bedrooms - days vacant <= 64.0 mae = 15326.334 samples = 9 value = 152814.999 824->825 828 2 bedrooms days vacant <= 8.0 mae = 17953.514 samples = 35 value = 130029.003 824->828 826 mae = 24710.751 samples = 4 value = 166052.5 825->826 827 mae = 7385.4 samples = 5 value = 150648.002 825->827 829 1 bedrooms days vacant <= 1.5 mae = 9656.901 samples = 20 value = 121435.5 828->829 834 2 bedrooms days vacant <= 64.5 mae = 20907.4 samples = 15 value = 144297.001 828->834 830 2 bedrooms days vacant <= 0.5 mae = 11027.923 samples = 13 value = 120250.002 829->830 833 mae = 6149.859 samples = 7 value = 126976.002 829->833 831 mae = 14535.144 samples = 7 value = 117289.0 830->831 832 mae = 6442.665 samples = 6 value = 121435.5 830->832 835 2 bedrooms days vacant <= 39.0 mae = 20826.545 samples = 11 value = 144687.998 834->835 838 mae = 15062.001 samples = 4 value = 128435.5 834->838 836 mae = 23066.6 samples = 5 value = 152312.998 835->836 837 mae = 17558.667 samples = 6 value = 143935.499 835->837 842 bldg_age <= 29.5 mae = 17293.15 samples = 28 value = 93709.151 841->842 851 0 bedrooms - days vacant <= 161.0 mae = 13894.069 samples = 39 value = 129231.083 841->851 843 0 bedrooms - days vacant <= 87.0 mae = 18282.467 samples = 21 value = 101690.001 842->843 850 mae = 4310.815 samples = 7 value = 81618.1 842->850 844 0 bedrooms - days vacant <= 61.0 mae = 14163.789 samples = 9 value = 106783.002 843->844 847 0 bedrooms - days vacant <= 118.0 mae = 19606.009 samples = 12 value = 92138.4 843->847 845 mae = 5004.819 samples = 5 value = 100545.999 844->845 846 mae = 16882.75 samples = 4 value = 122771.998 844->846 848 mae = 4104.659 samples = 5 value = 85066.8 847->848 849 mae = 23346.343 samples = 7 value = 105272.0 847->849 852 0 bedrooms - days vacant <= 76.5 mae = 12866.222 samples = 25 value = 139443.717 851->852 861 # of 0 bedrooms * days in month <= 2074.0 mae = 8733.906 samples = 14 value = 119426.941 851->861 853 # of days in month <= 30.5 mae = 8238.362 samples = 12 value = 128662.8 852->853 856 0 bedrooms - days vacant <= 101.5 mae = 10503.385 samples = 13 value = 150337.317 852->856 854 mae = 8458.783 samples = 7 value = 136236.055 853->854 855 mae = 4550.172 samples = 5 value = 124750.855 853->855 857 0 bedrooms - days vacant <= 84.0 mae = 6453.2 samples = 9 value = 151031.883 856->857 860 mae = 14722.0 samples = 4 value = 135440.941 856->860 858 mae = 3426.471 samples = 4 value = 155126.945 857->858 859 mae = 7520.8 samples = 5 value = 150037.145 857->859 862 mae = 2618.777 samples = 5 value = 116969.717 861->862 863 bldg_age <= 106.5 mae = 10934.724 samples = 9 value = 121983.255 861->863 864 mae = 5065.612 samples = 5 value = 122099.828 863->864 865 mae = 17729.057 samples = 4 value = 114977.8 863->865 867 0 bedrooms - days vacant <= 63.5 mae = 34492.998 samples = 50 value = 162630.501 866->867 884 bldg_age <= 102.5 mae = 40906.08 samples = 40 value = 136399.5 866->884 868 1 bedrooms days vacant <= 5.0 mae = 27567.011 samples = 35 value = 157946.0 867->868 879 0 bedrooms - days vacant <= 75.0 mae = 44681.367 samples = 15 value = 181543.997 867->879 869 0 bedrooms - days vacant <= 0.5 mae = 25290.321 samples = 28 value = 162630.501 868->869 878 mae = 22499.771 samples = 7 value = 129195.998 868->878 870 mae = 14855.249 samples = 4 value = 180444.499 869->870 871 0 bedrooms - days vacant <= 10.0 mae = 26340.666 samples = 24 value = 159515.5 869->871 872 mae = 8636.714 samples = 7 value = 150968.003 871->872 873 0 bedrooms - days vacant <= 44.5 mae = 32432.882 samples = 17 value = 163243.999 871->873 874 # of 0 bedrooms * days in month <= 1447.5 mae = 41220.398 samples = 10 value = 176978.5 873->874 877 mae = 13613.571 samples = 7 value = 157946.0 873->877 875 mae = 49976.666 samples = 6 value = 223499.998 874->875 876 mae = 13541.999 samples = 4 value = 161319.003 874->876 880 mae = 52077.644 samples = 7 value = 220626.998 879->880 881 0 bedrooms - days vacant <= 78.5 mae = 29281.5 samples = 8 value = 171983.5 879->881 882 mae = 26390.999 samples = 4 value = 171983.5 881->882 883 mae = 32172.001 samples = 4 value = 171754.497 881->883 885 bldg_age <= 101.5 mae = 23512.707 samples = 13 value = 95414.501 884->885 888 0 bedrooms - days vacant <= 156.0 mae = 37635.649 samples = 27 value = 143143.999 884->888 886 mae = 31214.251 samples = 6 value = 115702.999 885->886 887 mae = 5857.714 samples = 7 value = 82855.001 885->887 889 0 bedrooms - days vacant <= 112.0 mae = 29420.109 samples = 23 value = 143000.003 888->889 896 mae = 57765.001 samples = 4 value = 229720.0 888->896 890 # of days in month <= 30.5 mae = 29505.744 samples = 16 value = 145400.499 889->890 895 mae = 24835.658 samples = 7 value = 125019.0 889->895 891 # of 0 bedrooms * days in month <= 2146.0 mae = 23831.112 samples = 9 value = 143000.003 890->891 894 mae = 34436.414 samples = 7 value = 154774.998 890->894 892 mae = 24671.4 samples = 5 value = 144204.998 891->892 893 mae = 18461.501 samples = 4 value = 133983.499 891->893 898 0 bedrooms - days vacant <= 107.5 mae = 6270.195 samples = 17 value = 101154.0 897->898 903 bldg_age <= 115.5 mae = 16768.277 samples = 26 value = 109833.998 897->903 899 mae = 6818.286 samples = 7 value = 108834.0 898->899 900 # of 1 bedrooms * days in month <= 30.5 mae = 5118.53 samples = 10 value = 100295.951 898->900 901 mae = 4634.841 samples = 5 value = 93386.198 900->901 902 mae = 2806.22 samples = 5 value = 102338.002 900->902 904 0 bedrooms - days vacant <= 132.5 mae = 25446.022 samples = 9 value = 143312.002 903->904 907 0 bedrooms - days vacant <= 110.0 mae = 6010.53 samples = 17 value = 107502.002 903->907 905 mae = 18099.241 samples = 5 value = 139667.999 904->905 906 mae = 26310.5 samples = 4 value = 162933.501 904->906 908 0 bedrooms - days vacant <= 84.5 mae = 6286.077 samples = 13 value = 107550.001 907->908 913 mae = 4126.001 samples = 4 value = 102428.049 907->913 909 0 bedrooms - days vacant <= 56.0 mae = 5535.251 samples = 8 value = 107223.5 908->909 912 mae = 5172.801 samples = 5 value = 113159.0 908->912 910 mae = 4677.25 samples = 4 value = 107526.002 909->910 911 mae = 6114.75 samples = 4 value = 105746.499 909->911 915 1 bedrooms days vacant <= 0.5 mae = 17230.074 samples = 30 value = 73534.149 914->915 926 mae = 27959.499 samples = 4 value = 47499.998 914->926 916 mae = 3600.9 samples = 4 value = 60011.45 915->916 917 bldg_age <= 1.5 mae = 17772.146 samples = 26 value = 75558.801 915->917 918 # of days in month <= 30.5 mae = 10828.519 samples = 16 value = 72527.049 917->918 923 1 bedrooms days vacant <= 19.0 mae = 21940.61 samples = 10 value = 103861.499 917->923 919 mae = 8769.057 samples = 7 value = 62693.498 918->919 920 1 bedrooms days vacant <= 31.5 mae = 9703.023 samples = 9 value = 75000.001 918->920 921 mae = 15885.176 samples = 4 value = 75164.549 920->921 922 mae = 4757.301 samples = 5 value = 75000.001 920->922 924 mae = 20131.619 samples = 5 value = 104841.999 923->924 925 mae = 23357.401 samples = 5 value = 102881.0 923->925 928 0 bedrooms - days vacant <= 52.0 mae = 6325.649 samples = 36 value = 83549.26 927->928 943 mae = 2906.528 samples = 5 value = 73010.9 927->943 929 0 bedrooms - days vacant <= 19.5 mae = 4944.732 samples = 17 value = 82136.5 928->929 936 0 bedrooms - days vacant <= 97.0 mae = 6597.728 samples = 19 value = 86493.12 928->936 930 0 bedrooms - days vacant <= 6.5 mae = 3210.354 samples = 8 value = 83995.78 929->930 933 bldg_age <= 114.5 mae = 4810.663 samples = 9 value = 76606.24 929->933 931 mae = 1185.535 samples = 4 value = 82591.255 930->931 932 mae = 4082.273 samples = 4 value = 87513.65 930->932 934 mae = 2862.364 samples = 5 value = 78270.32 933->934 935 mae = 6693.682 samples = 4 value = 76031.315 933->935 937 0 bedrooms - days vacant <= 81.5 mae = 6832.0 samples = 14 value = 85396.645 936->937 942 mae = 4545.11 samples = 5 value = 89089.28 936->942 938 0 bedrooms - days vacant <= 68.0 mae = 7115.467 samples = 10 value = 87847.32 937->938 941 mae = 4533.673 samples = 4 value = 80342.185 937->941 939 mae = 5045.432 samples = 5 value = 86493.12 938->939 940 mae = 7205.564 samples = 5 value = 96392.81 938->940 945 bldg_age <= 7.5 mae = 17953.643 samples = 28 value = 200932.998 944->945 956 0 bedrooms - days vacant <= 588.5 mae = 29608.847 samples = 108 value = 158947.433 944->956 946 0 bedrooms - days vacant <= 71.0 mae = 16848.2 samples = 10 value = 213088.998 945->946 949 0 bedrooms - days vacant <= 37.5 mae = 13425.223 samples = 18 value = 190414.998 945->949 947 mae = 8505.666 samples = 6 value = 212127.499 946->947 948 mae = 26436.998 samples = 4 value = 238420.5 946->948 950 mae = 4174.501 samples = 4 value = 196681.002 949->950 951 # of days in month <= 30.5 mae = 15371.715 samples = 14 value = 189198.499 949->951 952 mae = 11892.666 samples = 6 value = 180755.498 951->952 953 bldg_age <= 8.5 mae = 16173.251 samples = 8 value = 190414.998 951->953 954 mae = 15084.001 samples = 4 value = 190414.998 953->954 955 mae = 17262.5 samples = 4 value = 193012.498 953->955 957 # of 3 bedrooms * days in month <= 88.5 mae = 24721.435 samples = 93 value = 155281.001 956->957 992 0 bedrooms - days vacant <= 741.5 mae = 34588.094 samples = 15 value = 216183.208 956->992 958 0 bedrooms - days vacant <= 1.0 mae = 19331.655 samples = 52 value = 149243.095 957->958 975 month_july <= 0.5 mae = 26531.463 samples = 41 value = 171639.0 957->975 959 mae = 7495.942 samples = 6 value = 130861.582 958->959 960 0 bedrooms - days vacant <= 179.0 mae = 19511.223 samples = 46 value = 150492.762 958->960 961 bldg_age <= 10.5 mae = 19148.757 samples = 40 value = 150992.878 960->961 974 mae = 14397.178 samples = 6 value = 126714.0 960->974 962 0 bedrooms - days vacant <= 134.5 mae = 11889.636 samples = 11 value = 159734.001 961->962 965 bldg_age <= 101.5 mae = 20496.122 samples = 29 value = 149170.19 961->965 963 mae = 5104.8 samples = 5 value = 162396.003 962->963 964 mae = 16816.0 samples = 6 value = 155590.001 962->964 966 0 bedrooms - days vacant <= 59.0 mae = 14843.006 samples = 21 value = 149315.999 965->966 971 0 bedrooms - days vacant <= 61.5 mae = 33670.264 samples = 8 value = 135281.283 965->971 967 0 bedrooms - days vacant <= 26.5 mae = 18982.602 samples = 14 value = 153438.226 966->967 970 mae = 3631.3 samples = 7 value = 145722.452 966->970 968 mae = 7532.014 samples = 7 value = 149170.19 967->968 969 mae = 22288.068 samples = 7 value = 175324.684 967->969 972 mae = 7952.241 samples = 4 value = 117396.184 971->972 973 mae = 36584.013 samples = 4 value = 162280.667 971->973 976 month_august <= 0.5 mae = 24524.757 samples = 37 value = 171510.998 975->976 991 mae = 6198.001 samples = 4 value = 215373.5 975->991 977 month_april <= 0.5 mae = 21592.576 samples = 33 value = 163510.002 976->977 990 mae = 6008.499 samples = 4 value = 214631.0 976->990 978 # of days in month <= 30.5 mae = 21560.607 samples = 28 value = 171511.498 977->978 989 mae = 9315.199 samples = 5 value = 145146.001 977->989 979 2 bedrooms days vacant <= 2.5 mae = 14748.556 samples = 9 value = 197876.002 978->979 982 month_may <= 0.5 mae = 17084.316 samples = 19 value = 162424.003 978->982 980 mae = 8562.4 samples = 5 value = 203002.0 979->980 981 mae = 13323.251 samples = 4 value = 178591.5 979->981 983 bldg_age <= 11.5 mae = 15133.933 samples = 15 value = 151178.999 982->983 988 mae = 13472.0 samples = 4 value = 176604.999 982->988 984 mae = 12291.714 samples = 7 value = 163510.002 983->984 985 bldg_age <= 12.5 mae = 14167.25 samples = 8 value = 146027.499 983->985 986 mae = 9205.5 samples = 4 value = 140721.999 985->986 987 mae = 17524.5 samples = 4 value = 161253.499 985->987 993 mae = 13622.154 samples = 5 value = 216651.821 992->993 994 0 bedrooms - days vacant <= 766.5 mae = 42184.196 samples = 10 value = 198850.006 992->994 995 mae = 22452.682 samples = 5 value = 195716.833 994->995 996 mae = 57684.526 samples = 5 value = 216872.756 994->996 998 # stories <= 6.0 mae = 38607.548 samples = 357 value = 190463.001 997->998 1135 hot water system_indirect hot water tank off boiler (heat & dhw) <= 0.5 mae = 60681.861 samples = 137 value = 272226.002 997->1135 999 heating system_boiler (hot water) <= 0.5 mae = 35822.892 samples = 268 value = 201774.497 998->999 1100 bldg_age <= 41.5 mae = 22715.0 samples = 89 value = 140531.002 998->1100 1000 2 bedrooms days vacant <= 20.5 mae = 23459.581 samples = 142 value = 188315.498 999->1000 1051 month_july <= 0.5 mae = 42049.23 samples = 126 value = 222371.0 999->1051 1001 bldg_age <= 10.5 mae = 23041.364 samples = 117 value = 191727.995 1000->1001 1042 # of 3 bedrooms * days in month <= 265.5 mae = 17689.277 samples = 25 value = 170499.997 1000->1042 1002 month_august <= 0.5 mae = 16952.726 samples = 74 value = 182303.401 1001->1002 1027 bldg_age <= 107.5 mae = 27499.909 samples = 43 value = 214225.001 1001->1027 1003 month_june <= 0.5 mae = 15863.236 samples = 69 value = 181670.32 1002->1003 1026 mae = 12510.973 samples = 5 value = 217053.891 1002->1026 1004 # of days in month <= 30.5 mae = 14729.518 samples = 62 value = 180426.38 1003->1004 1025 mae = 16615.784 samples = 7 value = 204782.995 1003->1025 1005 # of days in month <= 28.5 mae = 11613.929 samples = 24 value = 170903.379 1004->1005 1012 month_october <= 0.5 mae = 14848.901 samples = 38 value = 185670.597 1004->1012 1006 mae = 10271.149 samples = 6 value = 161865.502 1005->1006 1007 month_april <= 0.5 mae = 10995.83 samples = 18 value = 173564.496 1005->1007 1008 bldg_age <= 3.5 mae = 12701.687 samples = 12 value = 179538.377 1007->1008 1011 mae = 5164.117 samples = 6 value = 169382.002 1007->1011 1009 mae = 16434.799 samples = 5 value = 171847.995 1008->1009 1010 mae = 8835.784 samples = 7 value = 180243.756 1008->1010 1013 bldg_age <= 1.5 mae = 13518.962 samples = 31 value = 191390.998 1012->1013 1024 mae = 12458.738 samples = 7 value = 170499.997 1012->1024 1014 mae = 12399.8 samples = 5 value = 172763.998 1013->1014 1015 month_january <= 0.5 mae = 12991.839 samples = 26 value = 191559.497 1013->1015 1016 bldg_age <= 8.5 mae = 11941.225 samples = 19 value = 193553.991 1015->1016 1023 mae = 11935.288 samples = 7 value = 181670.32 1015->1023 1017 if multi-family sum of apt sq ft <= 43223.5 mae = 11983.918 samples = 14 value = 191559.497 1016->1017 1022 mae = 8918.919 samples = 5 value = 199819.966 1016->1022 1018 mae = 7618.03 samples = 4 value = 184831.462 1017->1018 1019 month_may <= 0.5 mae = 12787.3 samples = 10 value = 191727.995 1017->1019 1020 mae = 11989.497 samples = 6 value = 186000.002 1019->1020 1021 mae = 8087.002 samples = 4 value = 207397.002 1019->1021 1028 0 bedrooms - days vacant <= 0.5 mae = 29716.815 samples = 17 value = 251152.001 1027->1028 1033 1 bedrooms days vacant <= 41.5 mae = 14339.308 samples = 26 value = 196782.503 1027->1033 1029 # of 0 bedrooms * days in month <= 406.0 mae = 29772.936 samples = 12 value = 236306.501 1028->1029 1032 mae = 25454.72 samples = 5 value = 258099.034 1028->1032 1030 mae = 31992.181 samples = 5 value = 262958.311 1029->1030 1031 mae = 23237.43 samples = 7 value = 228305.999 1029->1031 1034 # of 1 bedrooms * days in month <= 2287.5 mae = 10190.251 samples = 8 value = 188329.497 1033->1034 1037 0 bedrooms - days vacant <= 30.5 mae = 13920.111 samples = 18 value = 201740.497 1033->1037 1035 mae = 11522.001 samples = 4 value = 188329.497 1034->1035 1036 mae = 8858.5 samples = 4 value = 187502.0 1034->1036 1038 1 bedrooms days vacant <= 63.0 mae = 9911.91 samples = 11 value = 195217.001 1037->1038 1041 mae = 13596.145 samples = 7 value = 218020.998 1037->1041 1039 mae = 10076.25 samples = 4 value = 190101.498 1038->1039 1040 mae = 9150.143 samples = 7 value = 198348.005 1038->1040 1043 bldg_age <= 109.5 mae = 15717.051 samples = 14 value = 153847.219 1042->1043 1048 2 bedrooms days vacant <= 30.5 mae = 11512.549 samples = 11 value = 181124.003 1042->1048 1044 0 bedrooms - days vacant <= 30.5 mae = 13648.673 samples = 10 value = 147913.504 1043->1044 1047 mae = 10707.001 samples = 4 value = 169535.5 1043->1047 1045 mae = 17971.184 samples = 4 value = 140230.859 1044->1045 1046 mae = 7660.238 samples = 6 value = 151543.003 1044->1046 1049 mae = 15109.181 samples = 4 value = 195256.269 1048->1049 1050 mae = 6426.317 samples = 7 value = 178613.504 1048->1050 1052 bldg_age <= 97.5 mae = 40421.244 samples = 115 value = 219155.001 1051->1052 1097 0 bedrooms - days vacant <= 87.0 mae = 42379.092 samples = 11 value = 269247.0 1051->1097 1053 # of 2 bedrooms * days in month <= 522.0 mae = 37033.269 samples = 104 value = 220009.504 1052->1053 1094 bldg_age <= 98.5 mae = 53769.182 samples = 11 value = 143505.003 1052->1094 1054 1 bedrooms days vacant <= 56.5 mae = 35130.11 samples = 27 value = 207408.001 1053->1054 1065 bldg_age <= 88.5 mae = 36771.351 samples = 77 value = 223551.002 1053->1065 1055 1 bedrooms days vacant <= 32.5 mae = 40295.764 samples = 17 value = 222292.002 1054->1055 1062 1 bedrooms days vacant <= 72.0 mae = 19673.299 samples = 10 value = 196396.5 1054->1062 1056 0 bedrooms - days vacant <= 138.5 mae = 34356.334 samples = 12 value = 220723.502 1055->1056 1061 mae = 49873.198 samples = 5 value = 245677.998 1055->1061 1057 month_february <= 0.5 mae = 39899.499 samples = 8 value = 222727.0 1056->1057 1060 mae = 17362.0 samples = 4 value = 198720.5 1056->1060 1058 mae = 28725.751 samples = 4 value = 234370.001 1057->1058 1059 mae = 49069.749 samples = 4 value = 202089.503 1057->1059 1063 mae = 6674.997 samples = 4 value = 206399.502 1062->1063 1064 mae = 24704.166 samples = 6 value = 193795.999 1062->1064 1066 month_august <= 0.5 mae = 31846.767 samples = 73 value = 222449.999 1065->1066 1093 mae = 30879.75 samples = 4 value = 365929.001 1065->1093 1067 0 bedrooms - days vacant <= 61.5 mae = 28427.123 samples = 65 value = 218463.001 1066->1067 1090 basement sq ft <= 3100.5 mae = 52561.998 samples = 8 value = 244294.001 1066->1090 1068 month_september <= 0.5 mae = 27483.88 samples = 58 value = 218273.999 1067->1068 1089 mae = 27363.141 samples = 7 value = 243579.002 1067->1089 1069 month_october <= 0.5 mae = 27710.34 samples = 53 value = 215672.001 1068->1069 1088 mae = 11868.201 samples = 5 value = 245534.002 1068->1088 1070 month_june <= 0.5 mae = 29069.501 samples = 48 value = 214186.499 1069->1070 1087 mae = 5921.799 samples = 5 value = 230641.004 1069->1087 1071 bldg_age <= 87.5 mae = 29525.81 samples = 42 value = 211896.503 1070->1071 1086 mae = 20614.001 samples = 6 value = 224753.503 1070->1086 1072 # of 0 bedrooms * days in month <= 1586.0 mae = 31891.136 samples = 37 value = 209967.999 1071->1072 1085 mae = 5279.199 samples = 5 value = 222449.999 1071->1085 1073 bldg_age <= 31.5 mae = 32540.94 samples = 33 value = 211492.006 1072->1073 1084 mae = 4335.25 samples = 4 value = 186699.999 1072->1084 1074 # of days in month <= 29.5 mae = 18117.77 samples = 26 value = 210532.501 1073->1074 1083 mae = 77180.429 samples = 7 value = 249110.004 1073->1083 1075 mae = 8413.747 samples = 4 value = 195827.5 1074->1075 1076 2 bedrooms days vacant <= 172.5 mae = 18644.591 samples = 22 value = 211896.503 1074->1076 1077 bldg_age <= 29.5 mae = 16601.389 samples = 18 value = 211294.504 1076->1077 1082 mae = 25749.001 samples = 4 value = 216968.501 1076->1082 1078 # of 2 bedrooms * days in month <= 701.5 mae = 20745.91 samples = 11 value = 212301.001 1077->1078 1081 mae = 7873.284 samples = 7 value = 203148.005 1077->1081 1079 mae = 15913.752 samples = 4 value = 216437.504 1078->1079 1080 mae = 23391.572 samples = 7 value = 211492.006 1078->1080 1091 mae = 63673.25 samples = 4 value = 273780.995 1090->1091 1092 mae = 20480.748 samples = 4 value = 224174.5 1090->1092 1095 mae = 42185.0 samples = 6 value = 138304.499 1094->1095 1096 mae = 42170.8 samples = 5 value = 235250.001 1094->1096 1098 mae = 34268.286 samples = 7 value = 251469.997 1097->1098 1099 mae = 42248.75 samples = 4 value = 291322.504 1097->1099 1101 bldg_age <= 3.5 mae = 17731.986 samples = 73 value = 135988.001 1100->1101 1128 # of days in month <= 30.5 mae = 14012.814 samples = 16 value = 189486.502 1100->1128 1102 1 bedrooms days vacant <= 23.5 mae = 12646.8 samples = 20 value = 127810.003 1101->1102 1109 month_august <= 0.5 mae = 18292.018 samples = 53 value = 139130.003 1101->1109 1103 # of 0 bedrooms * days in month <= 884.5 mae = 6611.8 samples = 15 value = 125261.002 1102->1103 1108 mae = 18643.999 samples = 5 value = 170865.999 1102->1108 1104 mae = 3654.143 samples = 7 value = 119273.0 1103->1104 1105 0 bedrooms - days vacant <= 6.0 mae = 6361.75 samples = 8 value = 129781.0 1103->1105 1106 mae = 3129.752 samples = 4 value = 126056.499 1105->1106 1107 mae = 6313.75 samples = 4 value = 134388.998 1105->1107 1110 bldg_age <= 5.5 mae = 17218.898 samples = 49 value = 138511.002 1109->1110 1127 mae = 20631.499 samples = 4 value = 171034.498 1109->1127 1111 1 bedrooms days vacant <= 5.0 mae = 9731.429 samples = 21 value = 146559.997 1110->1111 1118 bldg_age <= 39.5 mae = 20774.322 samples = 28 value = 133433.003 1110->1118 1112 bldg_age <= 4.5 mae = 5149.7 samples = 10 value = 148697.002 1111->1112 1115 0 bedrooms - days vacant <= 39.0 mae = 11474.091 samples = 11 value = 137591.997 1111->1115 1113 mae = 1710.001 samples = 4 value = 145245.497 1112->1113 1114 mae = 5829.498 samples = 6 value = 151602.998 1112->1114 1116 mae = 9255.001 samples = 5 value = 150705.001 1115->1116 1117 mae = 10948.501 samples = 6 value = 135534.999 1115->1117 1119 # of 2 bedrooms * days in month <= 377.0 mae = 10168.0 samples = 12 value = 121641.5 1118->1119 1122 bldg_age <= 40.5 mae = 23625.063 samples = 16 value = 142330.499 1118->1122 1120 mae = 9821.124 samples = 8 value = 116042.001 1119->1120 1121 mae = 1353.251 samples = 4 value = 127648.999 1119->1121 1123 mae = 25037.376 samples = 8 value = 166260.0 1122->1123 1124 1 bedrooms days vacant <= 101.0 mae = 9111.0 samples = 8 value = 134420.501 1122->1124 1125 mae = 11508.752 samples = 4 value = 137249.502 1124->1125 1126 mae = 4753.251 samples = 4 value = 129390.504 1124->1126 1129 1 bedrooms days vacant <= 148.5 mae = 9234.752 samples = 8 value = 186638.498 1128->1129 1132 2 bedrooms days vacant <= 9.0 mae = 17058.875 samples = 8 value = 196115.002 1128->1132 1130 mae = 13506.001 samples = 4 value = 189486.502 1129->1130 1131 mae = 2523.001 samples = 4 value = 183292.496 1129->1131 1133 mae = 14742.501 samples = 4 value = 196562.002 1132->1133 1134 mae = 19267.252 samples = 4 value = 191837.0 1132->1134 1136 total # bedrooms  in building <= 94.0 mae = 36507.831 samples = 89 value = 237019.998 1135->1136 1171 bldg_age <= 5.5 mae = 46313.375 samples = 48 value = 350541.5 1135->1171 1137 bldg_age <= 7.5 mae = 22834.907 samples = 43 value = 202448.0 1136->1137 1152 bldg_age <= 8.5 mae = 22757.869 samples = 46 value = 267529.497 1136->1152 1138 # of 3 bedrooms * days in month <= 183.0 mae = 18652.5 samples = 26 value = 188023.001 1137->1138 1147 bldg_age <= 8.5 mae = 12336.941 samples = 17 value = 225198.004 1137->1147 1139 bldg_age <= 6.5 mae = 15079.0 samples = 10 value = 170807.502 1138->1139 1142 2 bedrooms days vacant <= 7.0 mae = 14982.187 samples = 16 value = 197401.997 1138->1142 1140 mae = 16672.601 samples = 5 value = 162390.998 1139->1140 1141 mae = 10633.6 samples = 5 value = 176649.997 1139->1141 1143 mae = 10482.143 samples = 7 value = 210120.002 1142->1143 1144 2 bedrooms days vacant <= 31.5 mae = 12475.555 samples = 9 value = 189674.0 1142->1144 1145 mae = 9475.751 samples = 4 value = 177711.499 1144->1145 1146 mae = 10433.398 samples = 5 value = 196239.998 1144->1146 1148 2 bedrooms days vacant <= 44.0 mae = 11018.273 samples = 11 value = 222193.002 1147->1148 1151 mae = 7022.666 samples = 6 value = 235607.499 1147->1151 1149 mae = 6834.143 samples = 7 value = 222550.998 1148->1149 1150 mae = 10047.999 samples = 4 value = 203940.999 1148->1150 1153 month_july <= 0.5 mae = 19215.026 samples = 39 value = 262204.0 1152->1153 1170 mae = 12665.286 samples = 7 value = 301529.002 1152->1170 1154 2 bedrooms days vacant <= 5.0 mae = 16487.229 samples = 35 value = 256764.001 1153->1154 1169 mae = 19516.248 samples = 4 value = 287751.0 1153->1169 1155 1 bedrooms days vacant <= 13.0 mae = 13456.153 samples = 13 value = 272226.002 1154->1155 1160 bldg_age <= 5.5 mae = 15193.591 samples = 22 value = 250819.002 1154->1160 1156 bldg_age <= 6.5 mae = 12625.556 samples = 9 value = 277813.997 1155->1156 1159 mae = 7107.0 samples = 4 value = 258390.501 1155->1159 1157 mae = 6929.749 samples = 4 value = 272693.001 1156->1157 1158 mae = 9357.8 samples = 5 value = 289789.998 1156->1158 1161 2 bedrooms days vacant <= 72.5 mae = 20158.624 samples = 8 value = 243764.997 1160->1161 1164 bldg_age <= 6.5 mae = 10108.0 samples = 14 value = 255417.999 1160->1164 1162 mae = 5491.5 samples = 4 value = 240261.498 1161->1162 1163 mae = 34563.749 samples = 4 value = 253115.498 1161->1163 1165 mae = 6247.599 samples = 5 value = 262255.005 1164->1165 1166 2 bedrooms days vacant <= 34.0 mae = 10048.445 samples = 9 value = 250167.005 1164->1166 1167 mae = 14432.2 samples = 5 value = 256764.001 1166->1167 1168 mae = 2529.998 samples = 4 value = 248412.5 1166->1168 1172 2 bedrooms days vacant <= 1.5 mae = 41105.7 samples = 10 value = 391923.5 1171->1172 1175 0 bedrooms - days vacant <= 12.0 mae = 37652.447 samples = 38 value = 331587.0 1171->1175 1173 mae = 23496.5 samples = 6 value = 381962.5 1172->1173 1174 mae = 34547.5 samples = 4 value = 459344.5 1172->1174 1176 month_june <= 0.5 mae = 33581.467 samples = 30 value = 326315.0 1175->1176 1189 2 bedrooms days vacant <= 2.0 mae = 36530.375 samples = 8 value = 364453.0 1175->1189 1177 bldg_age <= 6.5 mae = 31098.231 samples = 26 value = 317041.5 1176->1177 1188 mae = 23893.0 samples = 4 value = 381709.5 1176->1188 1178 1 bedrooms days vacant <= 2.0 mae = 28293.0 samples = 9 value = 336305.0 1177->1178 1181 1 bedrooms days vacant <= 0.5 mae = 27205.529 samples = 17 value = 315528.0 1177->1181 1179 mae = 36848.6 samples = 5 value = 340064.0 1178->1179 1180 mae = 16658.75 samples = 4 value = 333957.0 1178->1180 1182 2 bedrooms days vacant <= 6.5 mae = 27248.692 samples = 13 value = 316426.0 1181->1182 1187 mae = 13346.75 samples = 4 value = 285528.5 1181->1187 1183 2 bedrooms days vacant <= 0.5 mae = 20589.889 samples = 9 value = 310032.0 1182->1183 1186 mae = 34768.5 samples = 4 value = 329859.5 1182->1186 1184 mae = 9487.0 samples = 4 value = 312948.0 1183->1184 1185 mae = 29472.2 samples = 5 value = 310032.0 1183->1185 1190 mae = 27829.75 samples = 4 value = 334656.5 1189->1190 1191 mae = 32924.0 samples = 4 value = 397170.0 1189->1191 1193 month_may <= 0.5 mae = 25526.98 samples = 51 value = 468904.997 1192->1193 1214 # of 2 bedrooms * days in month <= 1003.0 mae = 42569.862 samples = 94 value = 360938.505 1192->1214 1194 # of days in month <= 28.5 mae = 23129.934 samples = 46 value = 465587.006 1193->1194 1213 mae = 14909.602 samples = 5 value = 509286.008 1193->1213 1195 mae = 7779.499 samples = 4 value = 426174.498 1194->1195 1196 month_july <= 0.5 mae = 21440.499 samples = 42 value = 468188.998 1194->1196 1197 month_june <= 0.5 mae = 20862.21 samples = 38 value = 465587.006 1196->1197 1212 mae = 10312.75 samples = 4 value = 496027.497 1196->1212 1198 2 bedrooms days vacant <= 119.0 mae = 20087.97 samples = 34 value = 464362.499 1197->1198 1211 mae = 11112.75 samples = 4 value = 489093.004 1197->1211 1199 month_march <= 0.5 mae = 18861.333 samples = 30 value = 465015.501 1198->1199 1210 mae = 20760.25 samples = 4 value = 440406.499 1198->1210 1200 2 bedrooms days vacant <= 5.5 mae = 20033.519 samples = 25 value = 464044.003 1199->1200 1209 mae = 8857.0 samples = 5 value = 473220.997 1199->1209 1201 mae = 14643.001 samples = 4 value = 447088.008 1200->1201 1202 2 bedrooms days vacant <= 46.0 mae = 19818.048 samples = 21 value = 464680.995 1200->1202 1203 1 bedrooms days vacant <= 46.5 mae = 19361.616 samples = 13 value = 468904.997 1202->1203 1206 2 bedrooms days vacant <= 81.0 mae = 18600.999 samples = 8 value = 459566.003 1202->1206 1204 mae = 8666.669 samples = 6 value = 480296.498 1203->1204 1205 mae = 22654.428 samples = 7 value = 464680.995 1203->1205 1207 mae = 31628.001 samples = 4 value = 443406.501 1206->1207 1208 mae = 4563.0 samples = 4 value = 463044.002 1206->1208 1215 0 bedrooms - days vacant <= 0.5 mae = 31108.482 samples = 54 value = 344555.998 1214->1215 1236 bldg_age <= 10.5 mae = 46671.425 samples = 40 value = 400047.502 1214->1236 1216 1 bedrooms days vacant <= 0.5 mae = 28564.109 samples = 46 value = 337522.003 1215->1216 1233 1 bedrooms days vacant <= 17.0 mae = 27289.376 samples = 8 value = 393008.005 1215->1233 1217 # of days in month <= 30.5 mae = 20658.068 samples = 15 value = 351304.005 1216->1217 1222 month_march <= 0.5 mae = 28389.259 samples = 31 value = 327960.997 1216->1222 1218 mae = 23864.832 samples = 6 value = 332821.502 1217->1218 1219 bldg_age <= 12.5 mae = 15939.001 samples = 9 value = 361833.004 1217->1219 1220 mae = 24791.253 samples = 4 value = 352996.0 1219->1220 1221 mae = 8563.399 samples = 5 value = 363302.004 1219->1221 1223 1 bedrooms days vacant <= 4.5 mae = 27977.779 samples = 27 value = 328611.998 1222->1223 1232 mae = 21909.999 samples = 4 value = 309479.495 1222->1232 1224 mae = 30976.2 samples = 5 value = 339833.0 1223->1224 1225 1 bedrooms days vacant <= 19.5 mae = 26422.183 samples = 22 value = 327936.998 1223->1225 1226 # of 4 bedrooms * days in month <= 61.0 mae = 26638.8 samples = 10 value = 321605.498 1225->1226 1229 2 bedrooms days vacant <= 10.0 mae = 25668.334 samples = 12 value = 333059.506 1225->1229 1227 mae = 11770.801 samples = 5 value = 327467.998 1226->1227 1228 mae = 36030.6 samples = 5 value = 300086.995 1226->1228 1230 mae = 22366.0 samples = 7 value = 349474.996 1229->1230 1231 mae = 25979.202 samples = 5 value = 327912.998 1229->1231 1234 mae = 19885.749 samples = 4 value = 414072.003 1233->1234 1235 mae = 19101.004 samples = 4 value = 373836.002 1233->1235 1237 2 bedrooms days vacant <= 46.5 mae = 35728.666 samples = 9 value = 325955.995 1236->1237 1240 2 bedrooms days vacant <= 113.5 mae = 40605.581 samples = 31 value = 404376.005 1236->1240 1238 mae = 52349.0 samples = 5 value = 358202.995 1237->1238 1239 mae = 6891.498 samples = 4 value = 325039.997 1237->1239 1241 1 bedrooms days vacant <= 55.5 mae = 33307.92 samples = 25 value = 404280.994 1240->1241 1250 mae = 49049.669 samples = 6 value = 482218.999 1240->1250 1242 # of days in month <= 30.5 mae = 25514.264 samples = 19 value = 407603.995 1241->1242 1249 mae = 15910.666 samples = 6 value = 336862.001 1241->1249 1243 mae = 7611.287 samples = 7 value = 404280.994 1242->1243 1244 1 bedrooms days vacant <= 15.5 mae = 30791.752 samples = 12 value = 426757.5 1242->1244 1245 mae = 24596.752 samples = 4 value = 445300.999 1244->1245 1246 bldg_age <= 11.5 mae = 28575.5 samples = 8 value = 419566.5 1244->1246 1247 mae = 24698.254 samples = 4 value = 426757.5 1246->1247 1248 mae = 32013.252 samples = 4 value = 402100.502 1246->1248

Save the selected estimator


In [92]:
# Save the trained model
joblib.dump(clf2, os.path.join('model', 'trained_model.pkl'))


Out[92]:
['model/trained_model.pkl']

In [93]:
# Save the test data to a csv file
df_ = pd.DataFrame(data=X_valid, index=bldgs_valid, columns=features)
df_.to_csv(os.path.join('model', 'test_data.csv'), encoding='utf-8', index_label='buildings')

In [94]:
# These are the predictions on the data from the trained model...
p = pd.DataFrame(data=clf2.predict(X_valid), index=bldgs_valid, columns=['Predicted Monthly Water Consumption'])
p = pd.concat([p, df_], axis=1)
p.to_csv(os.path.join('model', 'predictions_test.csv'), encoding='utf-8', index_label='buildings')