I'm grappling with two parameters exhibiting an 82% correlation, showcasing a non-linear relationship and continuous fluctuations. Despite attempting LocalOutlierFactor in scikit-learn, outlier removal remains challenging. Parameter 1 ranges from 0.01 to 0.077, while Parameter 2 spans 3 to 12. Seeking advice on effective outlier removal methods to improve the correlation and better prepare the data for training a random forest regression model. Any insights or experiences would be greatly appreciated.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.