firmai/datagene ? reverse-engineered prompt

Build me a Python package called DataGene that helps compare time series datasets and tells me how similar they are.

I want to be able to load a few datasets, transform them into shapes that are easier to compare, and then get similarity scores and visual checks so I can see whether real and synthetic data, or train, test, and validation sets, actually match. It should work best with 3D time series data, but also handle 2D arrays when possible.

Please include the main transformation and comparison tools described in the project, things like PCA, SVD, recurrence plots, Gramian angular fields, Markov transition fields, correlation and covariance based views, plus distance and similarity measures for matrices and vectors. I’d also like simple plotting helpers for histograms, pairplots, correlation matrices, and the time series style visualizations.

Keep the package easy to import and use from notebooks, and make a small example showing how to compare a few datasets end to end. If you need to check current docs for any supporting libraries, go ahead and look them up online.