Is it possible to perform PCA with scikit learn or with other library in Python that uses Promax oblique rotation (with correlated principal components, not just ortogonal)? Does anybody tried this? This solution would be of great importance where latent structure of our features has some correlated components... For example, for this problem it would be very useful (IQ tests) https://www.kaggle.com/mpwolke/alpha-version-iq-test/notebook
Solution with two ortogonal (uncorrelated) components: https://github.com/Vitomir84/ML_algorithms/blob/main/PCA.ipynb