note

The effects of Kaizen can be predicted before implementation — Visualizing the "future difference" by integrating time-series forecasting and causal effects

This series, which began with 541,909 rows of purchase data, has finally reached its final installment. In the first installment, we viewed 4,338 customers as three "clusters" using RFM and K-Means.
Tech Times

LLM Data Mixture Breaks When Training Pools Shift: Causal Inference Offers Fix

LLM training data mixture optimization breaks when training pools shift — every prior proxy experiment becomes stale.