18 Libraries for Time Series Feature Extraction

Transforming Raw Time Series into Meaningful Features. The Data Scientist’s Toolkit for Time Series Feature Extraction

Photo by Mikael Kristenson on Unsplash

Feature extraction is a cornerstone step in many tasks involving time series. In this post, you’ll learn about 18 Python packages for extracting time series features.

Why extract time series features?

A feature is a statistic about a given time series, such as the mean value or seasonal strength.

Feature extraction provides a systematic way of describing time series, with important benefits. These include:

• Common representation: Different series may have a varied and potentially large number of points. Thus, summarizing them into a common representation enables dimensionality reduction and more meaningful comparisons.
• Interpretability: Features provide an efficient and interpretable way of summarizing the structure and dynamics of time series. Measures such as seasonal strength reveal quick and interpretable insights from time series that are not easy to get from raw data points.
• Performance: Well-designed features often boost the performance of algorithms in several tasks, including clustering…