05_independent_practice_solutions.ipynb

Kernel: Python 2

Time Series Independent Practice

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

%matplotlib inline

Walmart Sales Data

For this independent practice, we'll analyze Walmart's weekly sales data over a two-year period from 2010 to 2012.

The data set is again separated by store and department, but we'll focus on analyzing one store for simplicity.

The data include:

• Store: The store number.

• Dept: The department number.

• Date: The week.

• Weekly_Sales: Sales for the given department in the given store.

• IsHoliday: Whether the week is a special holiday week.

1) Preprocess the data using Pandas.

• Load the data.

• Convert the Date column to a datetime object.

• Set Date as the index of the DataFrame.

walmart = pd.read_csv('data/train.csv')

walmart.head()

walmart.dtypes

Store             int64
Dept              int64
Date             object
Weekly_Sales    float64
IsHoliday          bool
dtype: object

walmart['Date'] = pd.to_datetime(walmart['Date'])

walmart.dtypes

Store                    int64
Dept                     int64
Date            datetime64[ns]
Weekly_Sales           float64
IsHoliday                 bool
dtype: object

walmart.set_index('Date', inplace=True)

walmart.head()

2) Filter the DataFrame to Store 1 sales and aggregate over departments to compute the total weekly sales per store. Store this in a new DataFrame.

store1 = walmart[walmart.Store == 1][['Weekly_Sales']].resample('W', 'sum')

/Users/nicholebennett/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:1: FutureWarning: how in .resample() is deprecated
the new syntax is .resample(...).sum()
  """Entry point for launching an IPython kernel.

store1.head()

3) Plot the rolling mean for Weekly_Sales. What general trends do you observe?

store1[['Weekly_Sales']].rolling(window = 3, center = True).mean().plot()

<matplotlib.axes._subplots.AxesSubplot at 0x1c0dd14860>

4) Compute the 1, 13, and 52 autocorrelations for Weekly_Sales and/or create an autocorrelation plot.

print('Autocorrelation 1: ', store1['Weekly_Sales'].autocorr(1))
print('Autocorrelation 13: ', store1['Weekly_Sales'].autocorr(13))
print('Autocorrelation 52: ', store1['Weekly_Sales'].autocorr(52))

Autocorrelation 1:  0.302158279411
Autocorrelation 13:  0.10169228502
Autocorrelation 52:  0.895376029478

from pandas.tools.plotting import autocorrelation_plot

autocorrelation_plot(store1['Weekly_Sales'])

/Users/nicholebennett/anaconda3/lib/python3.6/site-packages/ipykernel_launcher.py:3: FutureWarning: 'pandas.tools.plotting.autocorrelation_plot' is deprecated, import 'pandas.plotting.autocorrelation_plot' instead.
  This is separate from the ipykernel package so we can avoid doing imports until

<matplotlib.axes._subplots.AxesSubplot at 0x10ac317b8>

from statsmodels.graphics.tsaplots import plot_acf

plot_acf(store1['Weekly_Sales'], lags=60)
plt.show()

5) Create a decomposition plot for the Store 1 sales data.

from statsmodels.tsa.seasonal import seasonal_decompose
decomposition = seasonal_decompose(store1.Weekly_Sales, freq=13)   
decomposition.plot() 
plt.show()

6) Based on the analyses above, what can we deduce about this time series?

# Big autocorrelation spikes are happening around 52, indicating some kind of yearly pattern.
# Autocorrelation is high at 1 and 2 lags (perhaps even up to 4), so it's most likely useful for an autoregressive model.
# There are no random spikes, so there's probably not much use for a moving average model.