The extraction method is a process in time series analysis where multiple constructed elements are used as input, and a singular value is extracted from each time series to create new augmented data. This method uses a package called TSfresh and includes default and custom features. The output of extraction methods differs from the output of transformation and interaction methods, as the latter outputs entirely new time series data. You can use this method when specific values need to be pulled from time series data.
The DeltaPy library contains 34 extraction methods. Writing the wrapper functions for extraction is similar to the wrapper transformation functions. The difficulty is how to discern the forecasting’s effectiveness from tabular augmentation. Furthermore, these methods are components and not complete functions for tabular augmentation.
Pluto will not explain each function, but here is a list of the extraction functions in the DeltaPy...