Linear Regression | What is multivariate normality in assumptions of Linear Regression?

Question

LR

in progress 0
TheDataMonk 4 years 2 Answers 1330 views Grand Master 0

Answers ( 2 )

    0

    Linear linear regression analysis requires all variables to be multivariate normal this can be best checked with the help of histogram for Q-Q plot(qqplot from statsmodel). I other words X and Y should be normally distributef.Normality can be checked using Kolmogrorov Smirnov test or simply by looking into the skewness and kurtosis for the given data. Also you can look into Shapiro Will test to know if the data was drawn from a Gaussian distribution sample. If the data is Not normally distributed you can use any non linear correction like log transform to make the data normally distributed

    0

    If the data points present on the plane which are normally disturbed along the regression line.If we plot histogram of residuals it will follow a normal distribution.

Leave an answer

Browse
Browse