Time series preparation functions for model training and applying
Model Applying function
Adjusted function for thresholds aggregation
Correction and Scale coefficients were integrated to the model applying procedure, so that it's possible to adjust thresholds without model re-training
Parameters for "Simple mode" were proposed:
High sensitivity
agg_th = 0.5, local_trend = 0.1, k = 0.1, similar = 0.01
Medium sensitivity
agg_th = 0.6, local_trend = 0.3, k = 0.1, similar = 0.1
Low sensitivity
agg_th = 0.8, local_trend = 0.5, k = 0.1, similar = 0.3