Repo structure

Folder: `data/`¶

Contains all datasets available for experiments. Metadata is stored in data/metadata.xlsx.

Includes all files required to build the documentation page on GitHub Desktop, including images.

Stores all experiment outputs generated by the PyNNLF tool.
Testing results are saved in experiment_result/Archive/Testing Result.

result_summary.xlsx: Summary of all experiment results, generated by notebooks/result_aggregation.ipynb.
testing_benchmark.xlsx: Benchmark results from three different computers.

Contains all source code for the tool.

Stores configuration files with hard-coded values.

config.ipynb: Contains tool parameters that rarely change (e.g., cross-validation folds, lookback period, train-test split ratio).
constant.ipynb: Stores constants like month names.
general_functions.ipynb: Core backend functions for PyNNLF, including lag feature creation, train-test splitting, cross-validation, evaluation, etc.
hyperparameters.ipynb: Stores all model hyperparameters.

Contains all model files and experiment notebooks.

Each model file follows the naming format [model_id]_[model_name].ipynb.

run_experiment.ipynb: Runs a single experiment.
run_experiment_batch.ipynb: Runs batch experiments across multiple forecast problems or model configurations.

result_aggregation.ipynb: Aggregates results from all experiments in experiment_result/ into result_summary.xlsx.

run_experiment_batch.ipynb: Also used for testing the tool.
Important to run full tests after major updates, especially those affecting the backend.