notebook.community

Edit and run 





In [1]:



    


import os

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns

from autolearn.autolearn import AutoLearn



    











In [2]:



    


%matplotlib inline
float_formatter = lambda x: "%.2f" % x
np.set_printoptions(formatter={'float_kind':float_formatter})
  • Add option to remove outliers from all columns of dataset?

Import Data





In [3]:



    


train_file = os.path.normpath(os.getcwd() + '/data/train.csv')
test_file = os.path.normpath(os.getcwd() + '/data/test.csv')
submission_folder = os.path.normpath(os.getcwd() + '/submissions/')



    











In [4]:



    


autolearn = AutoLearn(encode_categoricals=True, impute=True, onehot=True, standardize=False, decompose=False, target='SalePrice', id_col='Id', error_metric='rmsle', impute_strategy='median')
# autolearn = AutoLearn(encode_categoricals=True, impute=True, onehot=True, standardize=False, decompose=False, target='SalePrice', id_col='Id', error_metric='mse', impute_strategy='median')
training_data, X_train, y = autolearn.process_training_data(train_file)
training_data.head()



    


















    
Out[4]:










  
    

      
      Id
      MSSubClass
      MSZoning
      LotFrontage
      LotArea
      Street
      Alley
      LotShape
      LandContour
      Utilities
      ...
      PoolArea
      PoolQC
      Fence
      MiscFeature
      MiscVal
      MoSold
      YrSold
      SaleType
      SaleCondition
      SalePrice
    

  
  
    

      0
      1
      60
      3
      65.0
      8450
      1
      NaN
      3
      3
      0
      ...
      0
      NaN
      NaN
      NaN
      0
      2
      2008
      8
      4
      208500
    

    

      1
      2
      20
      3
      80.0
      9600
      1
      NaN
      3
      3
      0
      ...
      0
      NaN
      NaN
      NaN
      0
      5
      2007
      8
      4
      181500
    

    

      2
      3
      60
      3
      68.0
      11250
      1
      NaN
      0
      3
      0
      ...
      0
      NaN
      NaN
      NaN
      0
      9
      2008
      8
      4
      223500
    

    

      3
      4
      70
      3
      60.0
      9550
      1
      NaN
      0
      3
      0
      ...
      0
      NaN
      NaN
      NaN
      0
      2
      2006
      8
      0
      140000
    

    

      4
      5
      60
      3
      84.0
      14260
      1
      NaN
      0
      3
      0
      ...
      0
      NaN
      NaN
      NaN
      0
      12
      2008
      8
      4
      250000
    

  



5 rows × 81 columns



















In [5]:



    


print(training_data.shape)
print(X_train.shape)



    


















    






(1460, 81)
(1460, 133)

















In [6]:



    


test_data, X_test = autolearn.process_test_data(test_file)
test_data.head()



    


















    
Out[6]:










  
    

      
      Id
      MSSubClass
      MSZoning
      LotFrontage
      LotArea
      Street
      Alley
      LotShape
      LandContour
      Utilities
      ...
      ScreenPorch
      PoolArea
      PoolQC
      Fence
      MiscFeature
      MiscVal
      MoSold
      YrSold
      SaleType
      SaleCondition
    

  
  
    

      0
      1461
      20
      2
      80.0
      11622
      1
      NaN
      3
      3
      0
      ...
      120
      0
      NaN
      2
      NaN
      0
      6
      2010
      8
      4
    

    

      1
      1462
      20
      3
      81.0
      14267
      1
      NaN
      0
      3
      0
      ...
      0
      0
      NaN
      NaN
      0
      12500
      6
      2010
      8
      4
    

    

      2
      1463
      60
      3
      74.0
      13830
      1
      NaN
      0
      3
      0
      ...
      0
      0
      NaN
      2
      NaN
      0
      3
      2010
      8
      4
    

    

      3
      1464
      60
      3
      78.0
      9978
      1
      NaN
      0
      3
      0
      ...
      0
      0
      NaN
      NaN
      NaN
      0
      6
      2010
      8
      4
    

    

      4
      1465
      120
      3
      43.0
      5005
      1
      NaN
      0
      1
      0
      ...
      144
      0
      NaN
      NaN
      NaN
      0
      1
      2010
      8
      4
    

  



5 rows × 80 columns

Models





In [7]:



    


autolearn.train_all(X_train, y)



    


















    






100% (  7 of 7) |#########################| Elapsed Time: 0:00:02 Time: 0:00:02

















In [8]:



    


autolearn.predict_all(X_train)



    


















    






100% (  7 of 7) |#########################| Elapsed Time: 0:00:02 Time: 0:00:02

















In [9]:



    


autolearn.visualize_all(y)



    


















    






100% (  7 of 7) |#########################| Elapsed Time: 0:00:04 Time: 0:00:04










    

















    

















    

















    

















    

















    

















    
























In [10]:



    


autolearn.score_all(y)



    


















    






100% (  7 of 7) |#########################| Elapsed Time: 0:00:00 Time: 0:00:00

















In [11]:



    


autolearn.cross_validate_all(X_train, y)



    


















    






N/A% (0 of N/A) |                        | Elapsed Time: 0:00:00 ETA:  --:--:--/Users/Austin/anaconda/envs/kaggle/lib/python3.5/site-packages/sklearn/model_selection/_split.py:581: Warning: The least populated class in y has only 1 members, which is too few. The minimum number of groups for any class cannot be less than n_splits=3.
  % (min_groups, self.n_splits)), Warning)
 28% (  2 of 7) |#######                   | Elapsed Time: 0:00:02 ETA: 0:00:06/Users/Austin/projects/autolearn/autolearn/autolearn.py:223: RuntimeWarning: invalid value encountered in log1p
  return np.sqrt(np.mean(np.power(np.log1p(y_pred) - np.log1p(y), 2)))
 85% (  6 of 7) |######################    | Elapsed Time: 0:00:07 ETA: 0:00:01/Users/Austin/anaconda/envs/kaggle/lib/python3.5/site-packages/sklearn/model_selection/_split.py:581: Warning: The least populated class in y has only 1 members, which is too few. The minimum number of groups for any class cannot be less than n_splits=3.
  % (min_groups, self.n_splits)), Warning)
/Users/Austin/projects/autolearn/autolearn/autolearn.py:223: RuntimeWarning: invalid value encountered in log1p
  return np.sqrt(np.mean(np.power(np.log1p(y_pred) - np.log1p(y), 2)))
100% (  7 of 7) |#########################| Elapsed Time: 0:00:07 Time: 0:00:07

















In [12]:



    


results = autolearn.get_results()
results



    


















    
Out[12]:










  
    

      
      score
      cv
      variance
      parameters
    

  
  
    

      forest
      0.070293
      [0.193934693545, 0.145764425462, 0.154033980036]
      5.745238e+09
      (DecisionTreeRegressor(criterion='mse', max_de...
    

    

      bayes
      0.085183
      [0.332962710143, 0.212423251971, 0.224180451795]
      6.426012e+09
      GaussianNB(priors=None)
    

    

      boost
      0.089811
      [0.168805473538, 0.123528263815, 0.127427070759]
      5.859817e+09
      ([DecisionTreeRegressor(criterion='friedman_ms...
    

    

      bayes_ridge
      0.142163
      [0.180493099145, 0.15372542792, 0.172257400999]
      5.304830e+09
      BayesianRidge(alpha_1=1e-06, alpha_2=1e-06, co...
    

    

      ridge
      0.143071
      [0.183321595536, nan, 0.168031824507]
      5.448559e+09
      Ridge(alpha=1.0, copy_X=True, fit_intercept=Tr...
    

    

      lasso
      0.154537
      [nan, nan, 0.166198334168]
      5.500244e+09
      Lasso(alpha=1.0, copy_X=True, fit_intercept=Tr...
    

    

      linear
      0.155034
      [0.208602533925, nan, 0.166108985072]
      5.502522e+09
      LinearRegression(copy_X=True, fit_intercept=Tr...
    

  




















In [13]:



    


sns.stripplot(x=results.index, y='score', data=results, color='r', label='score');  # only really works if score is mse
sns.stripplot(x=results.index, y='variance', data=results, color='b', label='variance');
plt.xticks(rotation=45);



    


















    
























In [14]:



    


s = results.apply(lambda x: pd.Series(x['cv']),axis=1).stack().reset_index(level=1, drop=True)
s.name = 'cv'
results2 = results.drop(['score','variance','cv', 'parameters'], axis=1).join(s)
results2



    


















    
Out[14]:










  
    

      
      cv
    

  
  
    

      bayes
      0.332963
    

    

      bayes
      0.212423
    

    

      bayes
      0.224180
    

    

      bayes_ridge
      0.180493
    

    

      bayes_ridge
      0.153725
    

    

      bayes_ridge
      0.172257
    

    

      boost
      0.168805
    

    

      boost
      0.123528
    

    

      boost
      0.127427
    

    

      forest
      0.193935
    

    

      forest
      0.145764
    

    

      forest
      0.154034
    

    

      lasso
      0.166198
    

    

      linear
      0.208603
    

    

      linear
      0.166109
    

    

      ridge
      0.183322
    

    

      ridge
      0.168032
    

  




















In [15]:



    


sns.boxplot(x=results2.index, y='cv', data=results2, linewidth=2.5);

Linear Regression





In [16]:



    


model = autolearn.train(X_train, y, 'linear')
y_predictions = autolearn.predict(X_train, model)
score = autolearn.score(y, y_predictions)
autolearn.visualize(y, y_predictions)
print(score)



    


















    






0.155033780192

Logistic Regression





In [17]:



    


model = autolearn.train(X_train, y, 'logistic')
y_predictions = autolearn.predict(X_train, model)
score = autolearn.score(y, y_predictions)
autolearn.visualize(y, y_predictions)
print(score)



    


















    






0.047023527463

Ridge Regression





In [18]:



    


model = autolearn.train(X_train, y, 'ridge')
y_predictions = autolearn.predict(X_train, model)
score = autolearn.score(y, y_predictions)
autolearn.visualize(y, y_predictions)
print(score)



    


















    






0.143071036816

Lasso





In [19]:



    


model = autolearn.train(X_train, y, 'lasso')
y_predictions = autolearn.predict(X_train, model)
score = autolearn.score(y, y_predictions)
autolearn.visualize(y, y_predictions)
print(score)



    


















    






0.154536567815

Bayes





In [20]:



    


model = autolearn.train(X_train, y, 'bayes')
y_predictions = autolearn.predict(X_train, model)
score = autolearn.score(y, y_predictions)
autolearn.visualize(y, y_predictions)
print(score)



    


















    






0.0851826565966

Bayes Ridge





In [22]:



    


model = autolearn.train(X_train, y, 'bayes_ridge')
y_predictions = autolearn.predict(X_train, model)
score = autolearn.score(y, y_predictions)
autolearn.visualize(y, y_predictions)
print(score)



    


















    






0.142163076189

Forest





In [23]:



    


model = autolearn.train(X_train, y, 'forest')
y_predictions = autolearn.predict(X_train, model)
score = autolearn.score(y, y_predictions)
autolearn.visualize(y, y_predictions)
print(score)



    


















    






0.0714718256863










    
























In [ ]:

Content source: austinmcconnell/autolearn

Similar notebooks:

notebook.community | gallery | about