panelsplit: a tool for panel data analysis

panelsplit is a Python package designed to facilitate time series cross-validation when working with multiple entities (aka panel data). This tool is useful for handling panel data in various stages throughout the data pipeline, including feature engineering, hyper-parameter tuning, and model estimation.

Installation

panelsplit is tested for compatibility with python versions >= 3.11. You can install panelsplit using pip:

pip install panelsplit

---

Documentation

To read the documentation, visit here.

Example Usage

import pandas as pd
from panelsplit.cross_validation import PanelSplit

# Generate example data
num_countries = 2
years = range(2001, 2004)
num_years = len(years)

data_dict = {
    'country_id': [c for c in range(1, num_countries + 1) for _ in years],
    'year': [year for _ in range(num_countries) for year in years],
    'y': np.random.normal(0, 1, num_countries * num_years),
    'x1': np.random.normal(0, 1, num_countries * num_years),
    'x2': np.random.normal(0, 1, num_countries * num_years)
}

panel_data = pd.DataFrame(data_dict)
panel_split = PanelSplit(periods = panel_data.year, n_splits =2)

splits = panel_split.split()

for train_idx, test_idx in splits:
    print("Train:"); display(panel_data.loc[train_idx])
    print("Test:"); display(panel_data.loc[test_idx])

For more examples and detailed usage instructions, refer to the examples directory in this repository. Also feel free to check out an introductory article on panelsplit.

Background

Work on panelsplit started at EconAI in December 2023 and has been under active development since then.

Contributing

Contributions to panelsplit are welcome! If you encounter any issues or have suggestions for improvements, please feel free to open an issue or submit a pull request on GitHub.

License

This project is licensed under the MIT License - see the LICENSE file for details.