As far my knowledge, Linear Regression assumes that data or columns are normally distributed and doesn’t have multicollinearity amongs the features, But when I apply Shapiro test, it shows that none of the columns is normally distributed. Then how can we use Linear Regression for Boston House Price datasets
Asked
Active
Viewed 47 times
0
-
The regression *errors* should be normally distributed, but not necessarily the data used in the regression. Consider the case of fitting a sine wave, or an exponential data set. – James Phillips Dec 15 '18 at 04:24
-
also see https://stats.stackexchange.com/questions/12262/what-if-residuals-are-normally-distributed-but-y-is-not and https://stats.stackexchange.com/questions/75054/how-do-i-perform-a-regression-on-non-normal-data-which-remain-non-normal-when-tr – Glen_b Dec 15 '18 at 04:33