This is the source code of the paper "Causal Feature Selection Framework for Stable Soft Sensor Modeling based on Time-Delayed Cross Mapping"
If you find the code useful, please give a star :)
Chen, Shi-Shun, Xiao-Yang Li, and Enrico Zio. "Causal Feature Selection Framework for Stable Soft Sensor Modeling based on Time-Delayed Cross Mapping." Advanced Engineering Informatics (2026).
• Varying causal strength across time delay is considered for feature selection.
• State space reconstruction is employed to deal with interdependent variables.
• Time-delayed partial cross mapping is proposed for direct causal inference.
• Features are selected automatically via causal strength and validation performance.
• Features selected by direct causality can enhance model stability.
Paper link:
The folder 1. case_de contains the source code of the Debutanizer Column case.
In the folder, 1. causal inference contains the causal inference code using TDCCM and TDPCM; 2. soft sensor modelling contains the soft sensor modeling code using the results of TDCCM and TDPCM.