Covers the basics of Regression Techniques on the Big Market Dataset Download the Dataset from the link: https://datahack.analyticsvidhya.com/contest/practice-problem-big-mart-sales-iii/register
Regression Algorithms have varied characteristics and have relative mean squared errors with some working well on large datasets while others working well on smaller datasets and have their own cases of underfitting, overfitting and needs regularization on different levels.
Linear Regression Residual Plot:
Ridge Regression Residual Plot:
Ridge Regression Coefficients:
Lasso Regression Residual Plot:
Lasso Regression Coefficients:
Elastic Net Regression Residual Plot:
Elastic Net Regression Coefficients:
Clearly, Elastic Ridge Regression works the best in case of large datasets combining L1 and L2 Regularization and giving a suitable output with minimal error compared to other Regression Algorithms.