This repository contains the implementation of deep learning and hybrid ensemble models with SHAP-based explainability used in the research paper:
“Analysing agricultural distress in the eastern plateau of West Bengal's Rarh Region: integrating hybrid deep ensemble and GIS-based soft computing” Published in Discover Applied Sciences, Springer Nature, Open Access, Volume 7, Article 664 (2025), on 18 June 2025.
🔍 Purpose
The models were developed to predict agricultural distress zones in the Rarh Region of West Bengal, India, by integrating:
Hybrid Deep Ensemble Learning (CNN, LSTM, DFFNN)
SHAP (SHapley Additive exPlanations) for interpretability
GIS-based soft computing techniques for spatial analysis
⚙️ Features
Preprocessing of agricultural and socio-economic data
Training deep learning models (CNN, LSTM, DFFNN)
Evaluation of predictive performance (accuracy, loss)
SHAP-based feature importance and visualization (bar plots, beeswarm plots)
Integration with GIS-based outputs for mapping agricultural distress
This code demonstrates how explainable AI can be combined with spatial analysis to study The Spatial extension of Rural Agricultural vulnerability. Plain Text Citation: Chowdhury, G. (2025). Analysing agricultural distress in the eastern plateau of West Bengal's Rarh Region: integrating hybrid deep ensemble and GIS-based soft computing. Research, 7, Article 664. Published 18 June 2025.https://link.springer.com/article/10.1007/s42452-025-07244-2. BibTeX Citation: @article{Chowdhury2025AgriculturalDistress, author = {Chowdhury, Gopal and Saha, A.K.}, title = {Analysing agricultural distress in the eastern plateau of West Bengal's Rarh Region: integrating hybrid deep ensemble and GIS-based soft computing}, journal = {Research}, volume = {7}, pages = {Article 664}, year = {2025}, month = {June}, publisher = {Open Access}, doi = {[Insert DOI here if available](https://doi.org/10.1007/s42452-025-07244-2]}