Different type of segmentation algorithms for time series rather than Sliding Window with Encog

Question

I would like to know if Encog AI Framework supports different type of segmentation algorithms rather than commond sliding widow approach for time series forecasting. Eamonn Keogh's survey said that sliding window is the worst.

Here is the survey https://www.cs.rutgers.edu/~pazzani/Publications/survey.pdf

Let me know if Encog already supports different types of time series segmentation for ts forecasting.

Thanks

Answer 1

Encog only has built in support for sliding window time-series. However, there is nothing that stops you from implementing your own encoding of time-series data and passing it to the Encog models. Fundamentally all Encog models simply accept an input vector and produce an output vector. The paper you attached looks like a good summary. I will take a look for the next version of Encog, I've thought of adding additional built-in methods of encoding time series.