Unsupervised Analysis of Days of Week

In [1]:

    

%matplotlib inline
import matplotlib.pyplot as plt
plt.style.use('seaborn')
import pandas as pd
import numpy as np
from sklearn.decomposition import PCA
from sklearn.mixture import GaussianMixture


    

In [2]:

    

from jupyterworkflow.data import get_fremont_data
data = get_fremont_data()
pivoted = data.pivot_table('Total', index=data.index.time, columns=data.index.date)
pivoted.plot(legend=False, alpha=0.01) #we now have a line for each day


    

    
Out[2]:





<matplotlib.axes._subplots.AxesSubplot at 0x113eaf4e0>






    

In [3]:

    

X = pivoted.fillna(0).T.values
X.shape
#24 hours per 1631 days - T transposes


    

    
Out[3]:





(1641, 24)

In [4]:

    

X2 = PCA(2, svd_solver='full').fit_transform(X)
X2.shape


    

    
Out[4]:





(1641, 2)

In [5]:

    

plt.scatter(X2[:, 0], X2[:, 1]) 
#This shows there are two types of days (clustered)


    

    
Out[5]:





<matplotlib.collections.PathCollection at 0x11638b4e0>






    

In [6]:

    

gmm = GaussianMixture(2).fit(X)
labels = gmm.predict(X)


    

In [7]:

    

plt.scatter(X2[:, 0], X2[:, 1], c=labels, cmap='rainbow')
plt.colorbar()


    

    
Out[7]:





<matplotlib.colorbar.Colorbar at 0x1164fba20>






    

In [8]:

    

#Display two charts side by side
fig, ax = plt.subplots(1, 2, figsize=(14,6))

# Show cluster 0
pivoted.T[labels == 0].T.plot(legend=False, alpha=0.1, ax=ax[0]);
# Show cluster 1
pivoted.T[labels == 1].T.plot(legend=False, alpha=0.1, ax=ax[1]);

ax[0].set_title('Purple Cluster')
ax[1].set_title('Red Cluster')


    

    
Out[8]:





<matplotlib.text.Text at 0x1164f6358>






    

In [9]:

    

dayofweek = pd.DatetimeIndex(pivoted.columns).dayofweek


    

In [10]:

    

plt.scatter(X2[:, 0], X2[:, 1], c=dayofweek, cmap='rainbow')
plt.colorbar()


    

    
Out[10]:





<matplotlib.colorbar.Colorbar at 0x118eca978>






    

In [11]:

    

#Non weekends but in the "holidays" (non commute) cluster
#in seattle: 6/2 bad weather, people did not commute
dates  = pd.DatetimeIndex(pivoted.columns)
dates[(labels == 1) & (dayofweek < 5)]


    

    
Out[11]:





DatetimeIndex(['2012-11-22', '2012-11-23', '2012-12-24', '2012-12-25',
               '2013-01-01', '2013-05-27', '2013-07-04', '2013-07-05',
               '2013-09-02', '2013-11-28', '2013-11-29', '2013-12-20',
               '2013-12-24', '2013-12-25', '2014-01-01', '2014-04-23',
               '2014-05-26', '2014-07-04', '2014-09-01', '2014-11-27',
               '2014-11-28', '2014-12-24', '2014-12-25', '2014-12-26',
               '2015-01-01', '2015-05-25', '2015-07-03', '2015-09-07',
               '2015-11-26', '2015-11-27', '2015-12-24', '2015-12-25',
               '2016-01-01', '2016-05-30', '2016-07-04', '2016-09-05',
               '2016-11-24', '2016-11-25', '2016-12-26', '2017-01-02',
               '2017-02-06'],
              dtype='datetime64[ns]', freq=None)

In [ ]: