search

mljar / mljar-supervised

Python package for AutoML on Tabular Data with Feature Engineering, Hyper-Parameters Tuning, Explanations and Automatic Documentation
https://mljar.com
MIT License
2.99k stars 400 forks source link

Need mean_squared_log_error for eval_metric #611

Open dgdhruvs opened 1 year ago

dgdhruvs commented 1 year ago

Need mean_squared_log_error for regression problem.

pplonski commented 1 year ago

Hi @dgdhruvs,

You can pass custom evaluation metric as in this comment https://github.com/mljar/mljar-supervised/issues/390#issuecomment-830049603 or you can log transform target column and then just use MSE, btw MLJAR AutoML should automatically do log transformation on targer column if distribution is skewed.

There is more work to be done if you wan to have custom loss function - then you need to do some code changes in the package.

Please let me know if you need more help.

All the best! Piotr