The extreme value is obtained at the interval endpoint for convex function, and therefore the endpoint degree of a sample is measured by making the unstable cuts of all attributes according to the basic property.
More details for requirements file.
>>> git clone https://github.com/Zeroto521/compress.git >>> cd compress >>> python setup.py install
>>> from compress import Compression >>> model = Compression(data, labels) # Guess you have `data` which the shape is `(n, m)` and one column `labels` which the shape is `(n, 1)`. >>> model.fit() # Then use `fit()` function to fit model with data >>> data_new = model.compress(data, labels, k=0) # Then let use the `compress` to compress the data. `k` is the threshold to compress data >>> data_new # just show it
MIT License. @Zeroto521