Feature extraction is the practice of enhancing machine learning by finding characteristics in the data that help solve a particular problem. For time series data, feature extraction can be performed using various time series analysis and decomposition techniques.