pychow 0.1.0

A Python implementation of the Chow Test for structural breaks.

PyChow: A Python Implementation of the Chow Test

Overview

The Chow Test is used to determine if there is a structural break in a regression model, often applied in time series or panel data. This implementation provides a simple, reusable interface for performing the test using Python.

---

Formula

The Chow Test statistic is calculated as follows:

[ F = \frac{(RSS_{combined} - (RSS_1 + RSS_2)) / k}{(RSS_1 + RSS_2) / (N_1 + N_2 - 2k)} ]

Where:

• ( RSS_{combined} ): Residual sum of squares for the combined regression model.
• ( RSS_1, RSS_2 ): Residual sum of squares for the two separate regressions.
• ( k ): Number of parameters (including intercept) in the regression model.
• ( N_1, N_2 ): Number of observations in each subset.

The ( F )-statistic follows the ( F )-distribution with degrees of freedom:

• Numerator: ( k )
• Denominator: ( N_1 + N_2 - 2k )

---

Installation

To install PyChow:

pip install pychow

---

Usage

Example Code

import pandas as pd
import numpy as np
from pychow import ChowTest

# Example dataset
data = pd.DataFrame({
    'x': np.arange(1, 21),
    'y': np.concatenate([np.arange(1, 11), np.arange(21, 31)])
})

# Perform Chow Test
breakpoint = 10
result = ChowTest.chow_test(data, breakpoint, dependent_var='y', independent_vars=['x'])
print(result)

Output

The result will be a dictionary with the Chow Test statistic and the p-value:

{
    "Chow Test Statistic": 180.0,
    "P-value": 0.0
}

Parameters

• data: A Pandas DataFrame containing your dataset.
• breakpoint: The index at which the dataset is split into two parts.
• dependent_var: Name of the dependent variable (target column).
• independent_vars: List of names for independent variables (features).

---

Features

• Easy-to-use interface.
• Works with any dataset in Pandas DataFrame format.
• Outputs a dictionary with the test statistic and p-value.

---

Contributing

Feel free to fork the repository, submit pull requests, or open issues for feature requests and bugs.

---

License

MIT License