Open kokarevn opened 1 year ago
Ridge regression have shown better results than Lasso and was accepted as our main. Because the dataset have been compiled very carefully and the stationarization had heavily scaled the data the best alpha regularization parameter estimated by cross validation is very low (0.02).
Link to the code: https://colab.research.google.com/drive/1tmsa-w4ZdDBx4YoF5-hrPwkxPLgkFSqS?usp=sharing
1) Lasso regularization (L1 norm) 2) Ridge regularization (L2 norm)