Determine a generalizable dim reduc metric. Explained variance only works with (linear) PCA methods! So hopefully mutual info or reconstruction error are sufficient. Find the levels for those two metrics that correspond to 80% explained variance (or higher) for PCA.
Then combine all into a single function, with toggles for each. Ideally will only need to call one dim reduc metric...
Determine a generalizable dim reduc metric. Explained variance only works with (linear) PCA methods! So hopefully mutual info or reconstruction error are sufficient. Find the levels for those two metrics that correspond to 80% explained variance (or higher) for PCA.
Then combine all into a single function, with toggles for each. Ideally will only need to call one dim reduc metric...