In [1]:

    

import pandas as pd
import matplotlib.pyplot as plt

from sklearn.preprocessing import PolynomialFeatures
from sklearn.linear_model import LinearRegression
from sklearn.pipeline import Pipeline

import numpy as np

%matplotlib inline


    

In [2]:

    

data = pd.read_csv('bikes_train.csv') # , index_col=0 to use the datetime column as the DataFrame index
data.head()


    

    
Out[2]:






  

    

      

      
datetime
      
season
      
holiday
      
workingday
      
weather
      
temp
      
atemp
      
humidity
      
windspeed
      
casual
      
registered
      
count
    
  
  

    

      
0
      
2011-01-01 00:00:00
      
1
      
0
      
0
      
1
      
9.84
      
14.395
      
81
      
0.0
      
3
      
13
      
16
    
    

      
1
      
2011-01-01 01:00:00
      
1
      
0
      
0
      
1
      
9.02
      
13.635
      
80
      
0.0
      
8
      
32
      
40
    
    

      
2
      
2011-01-01 02:00:00
      
1
      
0
      
0
      
1
      
9.02
      
13.635
      
80
      
0.0
      
5
      
27
      
32
    
    

      
3
      
2011-01-01 03:00:00
      
1
      
0
      
0
      
1
      
9.84
      
14.395
      
75
      
0.0
      
3
      
10
      
13
    
    

      
4
      
2011-01-01 04:00:00
      
1
      
0
      
0
      
1
      
9.84
      
14.395
      
75
      
0.0
      
0
      
1
      
1
    
  

In [3]:

    

data.shape


    

    
Out[3]:





(10886, 12)

In [4]:

    

data.isnull().values.any()


    

    
Out[4]:





False

In [5]:

    

data.dropna()
data.shape


    

    
Out[5]:





(10886, 12)

In [6]:

    

def transform_data(data):
    data.datetime = data.datetime.apply(pd.to_datetime)
    data['month'] = data.datetime.apply(lambda x : x.month)
    data['hour'] = data.datetime.apply(lambda x : x.hour)
    data['day'] = data.datetime.apply(lambda x : x.timetuple().tm_yday)
    data['dayofweek'] = data.datetime.apply(lambda x : x.isoweekday())
    data.head()
    
transform_data(data)


    

In [7]:

    

from sklearn.model_selection import train_test_split
train_data, test_data = train_test_split(data)


    

In [8]:

    

fit_columns = ['season', 'holiday', 'workingday', 'weather', 'temp', 'atemp', 'humidity', 'windspeed', 'month', 'hour', 'dayofweek']


    

In [9]:

    

from sklearn.preprocessing import *

model = Pipeline([#('scaler', Normalizer()),
                  ('poly', PolynomialFeatures(degree=1)),
                  ('linear', LinearRegression(fit_intercept=False))])

model.fit(train_data[fit_columns], train_data[["count"]])


    

    
Out[9]:





Pipeline(steps=[('poly', PolynomialFeatures(degree=1, include_bias=True, interaction_only=False)), ('linear', LinearRegression(copy_X=True, fit_intercept=False, n_jobs=1, normalize=False))])

In [10]:

    

test_predictions = model.predict(test_data[fit_columns])
test_predictions.shape


    

    
Out[10]:





(2722, 1)

In [11]:

    

test_data["prediction"] = np.ravel(test_predictions)
test_data.head()


    

    




D:\lib\anaconda\lib\site-packages\ipykernel\__main__.py:1: 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
  if __name__ == '__main__':






    
Out[11]:






  

    

      

      
datetime
      
season
      
holiday
      
workingday
      
weather
      
temp
      
atemp
      
humidity
      
windspeed
      
casual
      
registered
      
count
      
month
      
hour
      
day
      
dayofweek
      
prediction
    
  
  

    

      
5574
      
2012-01-07 09:00:00
      
1
      
0
      
0
      
1
      
9.02
      
11.365
      
80
      
8.9981
      
14
      
116
      
130
      
1
      
9
      
7
      
6
      
7.381190
    
    

      
8285
      
2012-07-06 14:00:00
      
3
      
0
      
1
      
1
      
36.90
      
40.150
      
31
      
11.0014
      
91
      
184
      
275
      
7
      
14
      
188
      
5
      
400.870569
    
    

      
8875
      
2012-08-12 04:00:00
      
3
      
0
      
0
      
1
      
26.24
      
30.305
      
69
      
6.0032
      
2
      
8
      
10
      
8
      
4
      
225
      
7
      
170.834261
    
    

      
8375
      
2012-07-10 08:00:00
      
3
      
0
      
1
      
1
      
29.52
      
34.850
      
74
      
8.9981
      
34
      
615
      
649
      
7
      
8
      
192
      
2
      
207.195152
    
    

      
5780
      
2012-01-16 00:00:00
      
1
      
1
      
0
      
1
      
5.74
      
7.575
      
46
      
8.9981
      
2
      
23
      
25
      
1
      
0
      
16
      
1
      
-9.851459
    
  

In [12]:

    

print("The mean square error is %d" % np.mean((test_data["prediction"] - test_data["count"]) ** 2))


    

    




The mean square error is 20986

In [13]:

    

test_data[["day", "count", "prediction"]].groupby("day").aggregate(np.sum).plot(figsize=(20, 8))


    

    
Out[13]:





<matplotlib.axes._subplots.AxesSubplot at 0x9a3ae01e80>






    

In [14]:

    

test_data[["month", "count", "prediction"]].groupby("month").aggregate(np.sum).plot(figsize=(20, 8))


    

    
Out[14]:





<matplotlib.axes._subplots.AxesSubplot at 0x9a3af4df60>






    

In [15]:

    

test_data[["dayofweek", "count", "prediction"]].groupby("dayofweek").aggregate(np.sum).plot(figsize=(20, 8))


    

    
Out[15]:





<matplotlib.axes._subplots.AxesSubplot at 0x9a3b290358>