Demo.ipynb

Kernel: Python 2

import pandas as pd

data = pd.read_csv('../assets/dataset/rossmann.csv', skipinitialspace=True)

data['Date'] = pd.to_datetime(data['Date'])
data.set_index('Date', inplace=True)

data['Year'] = data.index.year
data['Month'] = data.index.month

store1_data = data[data.Store == 1]

data.head()

/home/user/anaconda3/envs/py2/lib/python2.7/site-packages/IPython/core/interactiveshell.py:2723: DtypeWarning: Columns (7) have mixed types. Specify dtype option on import or set low_memory=False.
  interactivity=interactivity, compiler=compiler, result=result)

data.index.year

array([2015, 2015, 2015, ..., 2013, 2013, 2013], dtype=int32)

import seaborn as sb
%matplotlib inline

data[['Sales']].resample('M').mean().sort_values(by='Sales')

store1_data.index

DatetimeIndex(['2015-07-31', '2015-07-30', '2015-07-29', '2015-07-28',
               '2015-07-27', '2015-07-26', '2015-07-25', '2015-07-24',
               '2015-07-23', '2015-07-22',
               ...
               '2013-01-10', '2013-01-09', '2013-01-08', '2013-01-07',
               '2013-01-06', '2013-01-05', '2013-01-04', '2013-01-03',
               '2013-01-02', '2013-01-01'],
              dtype='datetime64[ns]', name=u'Date', length=942, freq=None)

store1_data_2015 = store1_data['2015']
store1_data_2015[
    store1_data_2015.Open==1
][['Sales']].plot()

<matplotlib.axes._subplots.AxesSubplot at 0x7f79e826de10>

sb.factorplot(
    col='Open',
    hue='Promo',
    x='Month',
    y='Sales',
    data=store1_data, 
    kind='box'
)

<seaborn.axisgrid.FacetGrid at 0x7f79ff1c6e50>

sb.factorplot(
    col='Open',
    x='DayOfWeek',
    y='Customers',
    data=store1_data,
    kind='box',
    
)

<seaborn.axisgrid.FacetGrid at 0x7f79e6199ad0>

data[['Sales']].resample('D').mean().rolling(window=15, center=False).mean().diff(1).sort_values(by='Sales')

average_daily_customers = data[['Customers']].resample('D').mean()
average_daily_sales = data[['Sales', 'Open']].resample('D').mean()

average_daily_customers['DiffVsLastWeek'] = average_daily_customers.diff(periods=7)
average_daily_sales['DiffVsLastWeek'] = average_daily_sales[['Sales']].diff(periods=7)

average_daily_sales[average_daily_sales.Open == 1].sort_values(by='DiffVsLastWeek')

average_daily_sales = data[['Sales', 'Open']].resample('D').mean()

average_daily_sales['Sales'].autocorr(lag=1)

-0.02585827600638357

average_daily_sales['Sales'].autocorr(lag=30)

-0.12984245822530616

data['Sales'].resample('D').mean().expanding().sum()

Date
2013-01-01    8.728456e+01
2013-01-02    6.320315e+03
2013-01-03    1.201343e+04
2013-01-04    1.796765e+04
2013-01-05    2.330540e+04
2013-01-06    2.343446e+04
2013-01-07    3.314464e+04
2013-01-08    4.099167e+04
2013-01-09    4.793929e+04
2013-01-10    5.489130e+04
2013-01-11    6.210144e+04
2013-01-12    6.749829e+04
2013-01-13    6.762748e+04
2013-01-14    7.290711e+04
2013-01-15    7.785114e+04
2013-01-16    8.259825e+04
2013-01-17    8.742924e+04
2013-01-18    9.293329e+04
2013-01-19    9.832360e+04
2013-01-20    9.845956e+04
2013-01-21    1.063170e+05
2013-01-22    1.137792e+05
2013-01-23    1.207268e+05
2013-01-24    1.273993e+05
2013-01-25    1.343775e+05
2013-01-26    1.396231e+05
2013-01-27    1.397512e+05
2013-01-28    1.450236e+05
2013-01-29    1.499139e+05
2013-01-30    1.552141e+05
                  ...     
2015-07-02    5.272519e+06
2015-07-03    5.280170e+06
2015-07-04    5.285600e+06
2015-07-05    5.285841e+06
2015-07-06    5.292604e+06
2015-07-07    5.298337e+06
2015-07-08    5.304540e+06
2015-07-09    5.310762e+06
2015-07-10    5.317099e+06
2015-07-11    5.322715e+06
2015-07-12    5.322963e+06
2015-07-13    5.332686e+06
2015-07-14    5.341063e+06
2015-07-15    5.348916e+06
2015-07-16    5.356427e+06
2015-07-17    5.363556e+06
2015-07-18    5.369340e+06
2015-07-19    5.369608e+06
2015-07-20    5.376190e+06
2015-07-21    5.381891e+06
2015-07-22    5.387371e+06
2015-07-23    5.393368e+06
2015-07-24    5.399289e+06
2015-07-25    5.404689e+06
2015-07-26    5.404946e+06
2015-07-27    5.414549e+06
2015-07-28    5.422724e+06
2015-07-29    5.430347e+06
2015-07-30    5.438238e+06
2015-07-31    5.447305e+06
Freq: D, Name: Sales, dtype: float64

data['2013-01-01'].Sales.sum()

97235

total_daily_sales = data[['Sales']].resample('D').sum()

total_daily_sales.expanding().sum()['2014-12']

total_daily_sales.index

DatetimeIndex(['2013-01-01', '2013-01-02', '2013-01-03', '2013-01-04',
               '2013-01-05', '2013-01-06', '2013-01-07', '2013-01-08',
               '2013-01-09', '2013-01-10',
               ...
               '2015-07-22', '2015-07-23', '2015-07-24', '2015-07-25',
               '2015-07-26', '2015-07-27', '2015-07-28', '2015-07-29',
               '2015-07-30', '2015-07-31'],
              dtype='datetime64[ns]', name=u'Date', length=942, freq='D')