Timeline for Extremely high MSE/MAE for Ridge Regression(sklearn) when the label is directly calculated from the features
Current License: CC BY-SA 4.0
11 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 18, 2020 at 15:33 | vote | accept | RAbraham | ||
| Mar 18, 2020 at 14:23 | answer | added | Ben Reiniger♦ | timeline score: 2 | |
| Mar 18, 2020 at 11:58 | comment | added | RAbraham | @BenReiniger. would you care to put your above comment as the answer so that I can mark it as the answer, it helped me significantly. | |
| Mar 17, 2020 at 23:20 | comment | added | RAbraham | It had 18k bad rows :(. Please see edit2. And LinearRegression worked well. I have posted the metrics above. Thanks so much! | |
| Mar 17, 2020 at 23:18 | history | edited | RAbraham | CC BY-SA 4.0 |
discovered faulty rows :(
|
| Mar 17, 2020 at 22:13 | comment | added | Ben Reiniger♦ | The scores you report are based on the unscaled labels and predictions, so the MSE and MAE make sense and are not relatively large. But the R2 score is worryingly low; it may be only due to the regularization penalty: try LinearRegression? | |
| Mar 17, 2020 at 21:53 | comment | added | RAbraham | updated. I simplified the code. Thanks! | |
| Mar 17, 2020 at 21:52 | history | edited | RAbraham | CC BY-SA 4.0 |
simplified code and added info
|
| Mar 17, 2020 at 21:22 | comment | added | Ben Reiniger♦ | Could you provide the ridge regression's coefficients, and the scalers' mean/var? | |
| Mar 17, 2020 at 19:42 | answer | added | Oxbowerce | timeline score: 1 | |
| Mar 17, 2020 at 19:09 | history | asked | RAbraham | CC BY-SA 4.0 |