Multi-Series Forecasting at Scale with StatsForecast
summary
Retail demand pipelines use StatsForecast for fast statistical forecasts with rolling-origin cross-validation and conformal prediction intervals; model selection per series reduces error versus one-model-for-all. Benchmarks claim StatsForecast runs far faster than legacy libraries (20× pmdarima, 500× Prophet) but local fitting takes ~3 minutes on 8 cores, while a cloud foundational model TimeGPT2 returns forecasts in seconds and reports ~60% improvement over naive baselines. An agentic pipeline orchestrates preprocessing, forecasting, evaluation and works end to end.
AI-written summary · about
chapters (5)
- 0:00Multi-series forecasting intro
- 2:49StatsForecast library overview
- 12:56TimeGPT foundational model
- 24:06Nexla AI forecasting solution
- 29:31Q&A
related talks
42mRobust Automated Forecasting in Python and R
42mRobust Automated Forecasting in Python and R
19mForecasting for Everyone - Foundation Models and the Future of Time Series Prediction
29mProbabilistic demand forecasting at scale
28mDemand Forecasting: Accuracy, Stability, Automation
22mRonald Menich, Chief Data Scientist, Predictix, LLC
31mFrom stockouts to happy customers: Proven solutions for time series forecasting in retail
45mForecasting at Scale