1 Comment

This post doesn't say anything about identification technique. The referred article uses Broyden method and this is why it is slow. There are other approaches. One of them uses Kaczmarz method is much more effective https://www.sciencedirect.com/science/article/abs/pii/S0952197620303742

The coding example can be found in http://openkan.org/DLpiecewiseCPP.html

The dataset 10 000 records, 5 features needs only 0.1 second. MIT paper Python example needs 23 seconds for 2000 records and 2 features.

Expand full comment