Jan 17, 2022 · Artificial Intelligence

Detect Concept Drift Without Prior: Simple Supervised Method Using LSTM or LightGBM

This article explains a practical, supervised approach to concept‑drift detection by labeling historical data as 0 and new samples as 1, training an LSTM (or LightGBM) classifier, using high‑confidence predictions to flag distribution‑inconsistent samples, and includes code and real‑world examples.

LSTMLightGBMconcept drift

0 likes · 7 min read

Detect Concept Drift Without Prior: Simple Supervised Method Using LSTM or LightGBM

supervised detection

Detect Concept Drift Without Prior: Simple Supervised Method Using LSTM or LightGBM