Creating all in one script and pyinstaller specs for building single file app, fixing error in cross fold validation, adding separate kalman filters in traces, updating models

Adding already trained models on points and traces in H5 format, those models are now used in the correction phase

Fixing errors in the correction part, adding spline interpolation and Kalman filters for correction.

Fixing errors in correction with custom method and interpolation, adding vanilla lstm grid search results

Splitting correction in two phases: deleting pauses and second phase (deltas or interpolation). First phase completed and errors fixed

Adding correction using interpolation for comparison, first results of grid search (custom lstm), some backprop tests

Eliminating pauses in a trace, outliers are corected based on learned deltas. Grid Search code adapted and search started

Some more experiments on adversarial techniques for correction, fixing extra correction on already corrected points

Processed 26 traces, fixing evaluation plot and labels, testing the model on the new pre-processed data, running adversarial code

Split only on annotated points, criteria to decide if a segment needs to be discarded has been corrected, seaborn confusion matrix changed with color palettes reflecting numbers nature

Including all points in pre processing, sliding window chooses segments only containing annotated points. Output files updated

Ignoring points with missing altitude, make segments process now in a separate module, refreshing output object files