Panda > Statsmodel: syntax errors implementing variance_inflation_factor

Question

I am "using" Statsmodel for less than 2 days and am not at all familiar with the import commands etc. I want to run a simple variance_inflation_factor from here but am having some issues. My code follows:

from numpy import *
import numpy as np
import pandas as pd
from pandas import DataFrame, Series
import statsmodels.formula.api as sm
from sklearn.linear_model import LinearRegression
import scipy, scipy.stats
import matplotlib.pyplot as plt
import matplotlib
matplotlib.style.use('ggplot')
from statsmodels.api import add_constant
from numpy import linalg as LA
import statsmodels as sm

## I have been adding libraries and modules/packages with the intention of erring on the side of caution 

a = df1.years_exp
b = df1.leg_totalbills
c = df1.log_diff_rgdp
d = df1.unemployment
e = df1.expendituresfor
f = df1.direct_expenditures
g = df1.indirect_expenditures

sm.variance_inflation_factor((['a', 'b', 'c', 'd', 'e', 'f']), g)

then I get the following error:

AttributeError                            Traceback (most recent call last)
<ipython-input-61-bb126535eadd> in <module>()
----> 1 sm.variance_inflation_factor((['a', 'b', 'c', 'd', 'e', 'f']), g)

AttributeError: module 'statsmodels' has no attribute 'variance_inflation_factor'

Can someone direct me to the proper syntax for loading and executing this module? If it is more convenient that I post a link to some source code please ask. However, I have a feeling that this is just a simple syntax issue.

Answer 1

The function variance_inflation_factor is found in statsmodels.stats.outlier_influence as seen in the docs , so to use it you must import correctly, an option would be

from statsmodels.stats import outliers_influence
# code here 
outliers_influence.variance_inflation_factor((['a', 'b', 'c', 'd', 'e', 'f']), g)

Answer 2

Thanks for asking this question! I had the same question today, except I wanted to calculate the variance inflation factor for each of the features. Here is a programmatic way to do this:

from patsy import dmatrices
from statsmodels.stats.outliers_influence import variance_inflation_factor

# 'feature_1 + feature_2 ... feature_p'
features_formula = "+".join(df1.columns - ["indirect_expenditures"])

# get y and X dataframes based on this formula:
# indirect_expenditures ~ feature_1 + feature_2 ... feature_p
y, X = dmatrices('indirect_expenditures ~' + features_formula, df1, return_type='dataframe')

# For each Xi, calculate VIF and save in dataframe
vif = pd.DataFrame() 
vif["vif"] = [variance_inflation_factor(X.values, i) for i in range(X.shape[1])]
vif["features"] = X.columns
vif

Please note the above code works only if you have imported pandas and df1 is a pandas DataFrame

Answer 3

a = df1.years_exp
b = df1.leg_totalbills
c = df1.log_diff_rgdp
d = df1.unemployment
e = df1.expendituresfor
f = df1.direct_expenditures
g = df1.indirect_expenditures

ck=np.array([a,b,c,d,e,f,g])
outliers_influence.variance_inflation_factor(ck, 6)