Heteroscedasticity is an opposite concept to homoscedasticity. It is when there is unequal scatter of the residuals. This often occurs in situations with a large range present in the dataset. If there is heteroscedasticity present within the dataset, then the dataset is not eligible for linear regression.