Scientific Computing & Simulation / Independent exploration
IPS Time-series Experiments
A forecasting experiment harness comparing model wrappers across time-series datasets.
Problem
Forecasting comparisons need consistent data splits, horizons, metrics, and compute conditions.
Implementation
The inspected experiment runner registers PatchTST, TimesNet, NHITS, Prophet, and Chronos wrappers, sets NumPy and PyTorch seeds, and calls split/window utilities across a dataset registry.
Results
VERIFIED: model registration and experiment orchestration are implemented. This audit does not reproduce the model rankings.
Limitations
The runner contains absolute local dataset paths. Portable dataset provisioning, wrapper parity, and split consistency need verification before comparative performance claims.
Reproducibility
Reviewed at commit 8d340b45b824 on 2026-11-09. Source inspection establishes implementation structure; repository-authored results are not independent measurements. The upstream application and benchmark suite were not rerun during this review.