Replication package for "Does machine learning help manufacturers time base-metal purchases? Real-time, explainable evidence for six non-ferrous metals, 2012–2026"

Code and data to reproduce all tables and figures of the article. The package contains raw public data (IMF primary commodity prices and macro-financial series retrieved from FRED, the New York Fed Global Supply Chain Pressure Index; vintage of 23 October 2026), the assembled monthly panel (1998M01–2026M07), pseudo-out-of-sample forecasts of six LME base metals (copper, aluminium, nickel, zinc, lead, tin) at horizons of 1, 3 and 6 months from elastic net, random forest, XGBoost, forecast combination and econometric benchmarks, TreeSHAP attributions, procurement-timing decision outcomes, and the Python scripts that generate them. See README.md for instructions.Version: 1.1 (new version of the existing record if already published; title updated)