#10: Scikit-learn 7: Preprocessing 7: Intuition for Quantile Transform

[@antonelse]

Cool ! Thanks for the video ! Do you have a theoretical formulation of this procedure ? Or suggest one ? I appreciate your work man :)

[@learndataa]

Thank you. Hoping below helps: - Quantile: en.wikipedia.org/wiki/Quantile - Source code: github.com/scikit-learn/scikit-learn/blob/8c9c1f27b/sklearn/preprocessing/_data.py#L2382 - Quantile function: en.wikipedia.org/wiki/Quantile_function - Quantile transforms: scikit-learn.org/stable/modules/preprocessing.html#preprocessing-transformer - Example: scikit-learn.org/stable/auto_examples/compose/plot_transformed_target.html

[@vishalverma5837]

Great Video! I would like it more if you go a bit more in detail about the power transformer in order to build complete intuition.

[@kshitijdesai2402]

Since you mentioned Quantitle transformation can distort correlation and distances within and across features, then when to use this transformation and when not? What other alternative can be used to overcome this limitation?

[@ashokharnal8024]

Another reason for quantile transformation is data visualization. Clusters may be easily seen in quantile transformed data than without it.