How to use python to do regression with fixed effect and cluster

This blog mainly based on linearmodels (PyPI, Documents), which is a good interface to do regression analysis.

Sample data used in this article can be download here

Panel Models

Basic Linear Regression

First we load a sample dataset and use that dataset to do some simple regression

import numpy as np
from statsmodels.datasets import cpunish
data = cpunish.load_pandas().data

data.head()

	EXECUTIONS	INCOME	PERPOVERTY	PERBLACK	VC100k96	SOUTH	DEGREE
0	37.0	34453.0	16.7	12.2	644.0	1.0	0.16
1	9.0	41534.0	12.5	20.0	351.0	1.0	0.27
2	6.0	35802.0	10.6	11.2	591.0	0.0	0.21
3	4.0	26954.0	18.4	16.1	524.0	1.0	0.16
4	3.0	31468.0	14.8	25.9	565.0	1.0	0.19

import statsmodels.api as sm
dir(sm.datasets)

['__builtins__',
 '__cached__',
 '__doc__',
 '__file__',
 '__loader__',
 '__name__',
 '__package__',
 '__path__',
 '__spec__',
 'anes96',
 'cancer',
 'ccard',
 'check_internet',
 'china_smoking',
 'clear_data_home',
 'co2',
 'committee',
 'copper',
 'cpunish',
 'elnino',
 'engel',
 'fair',
 'fertility',
 'get_data_home',
 'get_rdataset',
 'grunfeld',
 'heart',
 'longley',
 'macrodata',
 'modechoice',
 'nile',
 'randhie',
 'scotland',
 'spector',
 'stackloss',
 'star98',
 'statecrime',
 'strikes',
 'sunspots',
 'utils',
 'webuse']

Written on April 23, 2018