Python – Statsmodels Example – Video

Statsmodels is a nice statistics library in Python, which eases the data processing and analysis with Python. The library is available here – statsmodels.org and on its official web site the installation guide is always up to date.

In the live coding video, the following points are mentioned:

• imporing libraries and loading macrodata dataset to pandas
• displaying data from dataset in all columns and all rows
• inserting index to the data
• plotting the data with matplotlib
• getting the trend and the cycle of the data with hp_filter.hpfilter (https://www.statsmodels.org/dev/generated/statsmodels.tsa.filters.hp_filter.hpfilter.html)
• presenting the data, zooming to specific dates

Imporing libraries and loading macrodata dataset to pandas is the standard start.

import pandas as pd
import matplotlib.pyplot as plt
import statsmodels.api as sm
%matplotlib inline
df = sm.datasets.macrodata.load_pandas().data

Displaying data from dataset in all columns and all rows is actually a 1-liner. 1 per column, 1 per row:

pd.options.display.max_columns = None 
pd.options.display.max_rows = None

Inserting the index to the data is a standard trick, which could be carried out through different ways – e.g. via length or via from-to period:

index = pd.Index(sm.tsa.datetools.dates_from_range('1959Q1', length=203))

index = pd.Index(sm.tsa.datetools.dates_from_range('1959Q1', '2009Q3'))
df.index = index

df[['trend', 'cycle']].plot()

Getting the trend and the cycile of the dataset with hp_filter.hpfilter is actually parsing data from a tuple to the dataset:

cycle, trend = sm.tsa.filters.hpfilter(df["realgdp"])
df['trend'] = trend
df['cycle'] = cycle

df[['trend', 'cycle']]["1985":"1999"].plot()