Get SCM Predictions and LOIO Probabilities

[roshankern]

This PR is ready for review!

In this PR, predictions and leave-one-image-out probabilities are derived for the single-class models.

For predictions, we simply load in the model for a particular phenotypic class, model type (final or shuffled baseline), and feature type (CP, DP, CP_and_DP) then derive model predictions for cells from both data subsets (train and test). Predictions are saved to predictions/.

The LOIO evaluation procedure is as follows:

• Load in entire MitoCheck labeled cell dataset (from labeled_data.csv.gz).
• For each image in MitoCheck labeled cell dataset (as specified by the Metadata_DNA field):
  • Train a logistic regression model with optimal hyperparameters (C and l1_ratio, saved with model data in models/) on every cell that is not in the specific image.
  • Predict probabilities on every cell that is in the specific image.

The probabilities are saved to LOIO_probas/.

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[roshankern]

Merging into wrong branch.