How to create Contingency tables in the StatsModels library?

Recipe Objective - How to create Contingency tables in the StatsModels library?

The statsmodels support a variety of approaches for analyzing contingency tables, including how to evaluate independence, symmetry, and uniformity, and how to work with collections of tables from a stratified population. The method described here applies primarily to bidirectional tables. Reusable tables can be analyzed with a log-linear model. There is currently no dedicated API for log-linear modeling in statsmodels, but you can use Poisson regression in statsmodels.genmod.GLM for this purpose.

Explore the Must Know Python Libraries for Data Science and Machine Learning.  

For more related projects -

https://www.dezyre.com/projects/data-science-projects/deep-learning-projects
https://www.dezyre.com/projects/data-science-projects/neural-network-projects

The statsmodels.stats.The table is the most basic class for working with contingency tables. You can create a Table object directly from an object like a rectangular array containing the cell numbers of a contingency table.

# Importing libraries
import statsmodels.api as sm
import pandas as pd

# Importing flchain dataset from survival package
X = sm.datasets.get_rdataset("flchain", "survival").data

# Crosstab of two features (chapter and sex)
crosstab = pd.crosstab(X['chapter'], X['sex'])

# Table
cont_table = sm.stats.Table(crosstab)
cont_table.table_orig


Output-
sex	F	M
chapter		
Blood	1	3
Circulatory	401	344
Congenital	0	3
Digestive	37	29
Endocrine	25	23
External Causes	35	31
Genitourinary	20	22
Ill Defined	25	13
Infectious	23	9
Injury and Poisoning	12	9
Mental	100	44
Musculoskeletal	11	3
Neoplasms	279	288
Nervous	73	57
Respiratory	121	124
Skin	2	2

In this way, we can create contingency tables in the StatsModels library.