# License scopes

Code, tests, browser UI and my documentation: Apache-2.0, copyright 2026 Renchao Wu. See LICENSE. Point-prediction helpers derive from my prior accepted local Bike Hour Forecast code; the exact prior core hash is recorded in RESEARCH.json. The original package was not edited.

Original UCI Bike Sharing archive/CSVs/README: Hadi Fanaee-T (2013), DOI 10.24432/C5W894, CC BY 4.0 as stated on the original UCI page. Original public contacts remain original source metadata. See data/SOURCE.json and data/LICENSE-CC-BY-4.0.

Derived trained weights, calibration residuals, experimental outputs, example and figures: CC BY 4.0 with the same dataset attribution. Original bytes are unchanged; transformations are described in PROTOCOL/README. No endorsement is asserted.

The method paper is linked/cited; its code, prose and figures are not bundled or relicensed. Mathematical rank tests and my implementation are separate from that paper's empirical studies.

Arbitrary user histories/scores retain their original terms. Dataset training-source attribution does not authenticate arbitrary user input. Python/NumPy/Node/Matplotlib are independently distributed dependencies with their respective licenses and are not bundled here.
