Online models were lacking in two capabilities: feature queueing for time series models (i.e., passing in an array of window x sequence x features) and feature standardization.
Added a parameter to the constructor that maintains a queue to pass a sequence of previous windows to time series models. Added an install_standardization method so users can pass in a StandardScaler or some training data to calculate means/standard deviations for standardizing features.
Online models were lacking in two capabilities: feature queueing for time series models (i.e., passing in an array of window x sequence x features) and feature standardization.
Added a parameter to the constructor that maintains a queue to pass a sequence of previous windows to time series models. Added an
install_standardizationmethod so users can pass in aStandardScaleror some training data to calculate means/standard deviations for standardizing features.TODO: