Currently, the raw data is interpolated to unify the sample rate of all timeSeries in the training set. The model-generation does not obey to this strategy. Either the model generation must know about the new, unified sample rate or another strategy has to be chosen.
If the sample rates for different sensors in the training set is different, sample based windows and step sizes make no sense.
Currently, the raw data is interpolated to unify the sample rate of all timeSeries in the training set. The model-generation does not obey to this strategy. Either the model generation must know about the new, unified sample rate or another strategy has to be chosen.
If the sample rates for different sensors in the training set is different, sample based windows and step sizes make no sense.