Optimizing Customer Risk Assessment with Tiered Models
Optimizing Customer Risk Assessment with Tiered Models
Created using ChatSlide
This project involves a comprehensive approach to detecting electricity theft using smart-meter data. It begins with data preparation and feature extraction, where SGCC data is cleaned and relevant features are extracted. Next, a LightGBM model is employed for Tier 1 screening to rank customers based on theft risk, followed by evaluation metrics like Precision@K and Recall@K. Finally, Tier 2 refinement utilizes Bi-LSTM for analyzing temporal patterns in selected customer sequences,...