TODO

Backlog:
-----------

Small:
** [Done] raise insights to user when there are important insights he should look into
** [Done] provide calibration curve for uncertainty models
** make separation quality into a report, and add sensitivity analysis for the separation classifier, and connect it to insights.
** add cross-report insights - e.g. features sensitive to missing values that also has many missing values
** add uncertainty of model to uncertainty predictions
** analyze which areas of the feature space is more prone to errors (either using uncertainty estimator, or basic analysis)
** Metrics - micro / binary / additional
** Uncertainty methods - prediction intervals regression - xgboost quantiles
** add meta-data to dataset analysis report (feature types, names, base types, etc.)

Big:
** Readable report (e.g. pdf)
** UI - widgets for notebooks
** Explainability   - shap / DT / KNN / linear
** Dataset quality  - values distribution in feature, recognize complex feature types
                    - Check if more data is required, Find areas with not enough samples
                    - Find anomalies / outliers in data
                    - Find anomalies / outliers in target
** Uncertainty methods - sample out of distribution


** Advanced ML - group id, sample weight
** problem types - vision / time series / anomaly detection