overfitting

generally;

• more data,
• cross-validation,
• reducing variables (future selection & extraction),
• early stop for dl, neural networks

                           overfitting                            scaling         multi-collinearity
         regression |   L1 (lasso), L2 (rigde) Regularization  | standardize  |  eliminate high correlated features or 
                                                                              |  use Ridge(Lasso)Regression which penalties high correlation
         trees      |   ensembling (bagging, boosting, RF)     | no-need      |    n/a
 neural network, DL |   early stop, drop-out                   | normalize    |    n/a