How to using co-variance for feature engineering
You could try PCA.
" Principal component analysis ( PCA) is a statistical procedure that uses an orthogonal transformation to convert a set of observations of possibly correlated variables (entities each of which takes on various numerical values) into a set of values of linearly uncorrelated variables called principal components."
Heres a quick tutorial for python implementation
Yeah I know this approach but I don’t want to use PCA or SVD or already built in techniques
Hmm I’m not sure exactly. You could make some polynomial features on the negative co-eff?