In [26]:

    

%matplotlib inline

import pandas as pd
import numpy as np

from fbprophet import Prophet


    

In [27]:

    

df = pd.read_csv('analytics-2017.pycon.ca-audience-overview_20170123-20170526.csv',
                 sep='\t', parse_dates=['Date'])
df.tail()


    

    
Out[27]:






  

    

      

      
Date
      
Sessions
    
  
  

    

      
119
      
2017-05-22
      
16
    
    

      
120
      
2017-05-23
      
31
    
    

      
121
      
2017-05-24
      
31
    
    

      
122
      
2017-05-25
      
25
    
    

      
123
      
2017-05-26
      
17
    
  

In [28]:

    

df.set_index('Date').plot()


    

    
Out[28]:





<matplotlib.axes._subplots.AxesSubplot at 0x10abd4278>






    

In [29]:

    

df['Sessions'] = np.log(df['Sessions'])
df.set_index('Date').plot()


    

    
Out[29]:





<matplotlib.axes._subplots.AxesSubplot at 0x10b01feb8>






    

In [30]:

    

df.columns = ['ds', 'y']
df.head()


    

    
Out[30]:






  

    

      

      
ds
      
y
    
  
  

    

      
0
      
2017-01-23
      
3.218876
    
    

      
1
      
2017-01-24
      
3.526361
    
    

      
2
      
2017-01-25
      
3.218876
    
    

      
3
      
2017-01-26
      
2.995732
    
    

      
4
      
2017-01-27
      
2.995732
    
  

In [31]:

    

prophet = Prophet()
prophet.fit(df)


    

    
Out[31]:





<fbprophet.forecaster.Prophet at 0x10ac89a20>

In [32]:

    

future = prophet.make_future_dataframe(periods=365)


    

In [33]:

    

forecast = prophet.predict(future)


    

In [34]:

    

forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].tail()


    

    
Out[34]:






  

    

      

      
ds
      
yhat
      
yhat_lower
      
yhat_upper
    
  
  

    

      
484
      
2018-05-22
      
2.056819
      
1.587713
      
2.512842
    
    

      
485
      
2018-05-23
      
2.126686
      
1.630847
      
2.562205
    
    

      
486
      
2018-05-24
      
1.796638
      
1.316910
      
2.316290
    
    

      
487
      
2018-05-25
      
1.842713
      
1.326227
      
2.302523
    
    

      
488
      
2018-05-26
      
1.378937
      
0.870114
      
1.867801
    
  

In [35]:

    

np.exp(forecast[['yhat', 'yhat_lower', 'yhat_upper']].tail())


    

    
Out[35]:






  

    

      

      
yhat
      
yhat_lower
      
yhat_upper
    
  
  

    

      
484
      
7.821049
      
4.892546
      
12.339945
    
    

      
485
      
8.387030
      
5.108202
      
12.964368
    
    

      
486
      
6.029342
      
3.731872
      
10.137993
    
    

      
487
      
6.313646
      
3.766805
      
9.999379
    
    

      
488
      
3.970677
      
2.387183
      
6.474047
    
  

In [36]:

    

prophet.plot(forecast)


    

    
Out[36]:












    

In [37]:

    

prophet.plot_components(forecast)


    

    
Out[37]:












    

In [ ]: