Econometron: A Python package for Econometric and Time Series Analysis

Introduction

Econometron is a Python library designed for econometric modeling, time series analysis, and dynamic stochastic general equilibrium (DSGE) model solving and estimation. It provides a wide range of tools for researchers, economists, and data scientists to build, estimate, and analyze complex multivariate time series models and non-linear DSGE models. With a focus on flexibility and performance, Econometron supports both classical and modern approaches, including state-of-the-art neural network-based forecasting and robust statistical methods.

Whether you're modeling economic time series, performing impulse response function (IRF) analysis, or solving non-linear DSGE models, Econometron offers a unified and efficient framework to streamline your workflow.

Key Features

Multivariate Statistical Time Series Models

• VAR (Vector Autoregression): Model the dynamic relationships between multiple time series.
• SVAR (Structural Vector Autoregression): Incorporate structural restrictions for causal inference and policy analysis.
• VARMA (Vector Autoregressive Moving Average): Combine autoregressive and moving average components for enhanced flexibility.
• VARIMA (Vector Autoregressive Integrated Moving Average): Handle non-stationary time series with differencing.

VARMA Identification

• Echelon Form Identification: Implements the echelon form approach for identifying VARMA models, ensuring robust and unique parameter estimation in Python.

Neural Network-Based Forecasting

• N-BEATS (Neural Basis Expansion Analysis for Time Series): A state-of-the-art deep learning model for univariate and multivariate time series forecasting.
• N-BEATS + RevIN: Enhances N-BEATS with Reversible Instance Normalization (RevIN) for improved generalization and robustness.

State Space Models

• Flexible framework for modeling complex dynamic systems using state space representations, suitable for both linear and non-linear systems.

Estimation Methods

• Bayesian Estimation: Leverage Bayesian techniques for parameter estimation, incorporating prior knowledge and uncertainty quantification.
• Maximum Likelihood Estimation (MLE): Optimize model parameters using likelihood-based methods for precise inference.

Impulse Response Functions (IRF)

• Local Projection IRF: Compute impulse response functions using local projection methods, ideal for non-linear and robust analysis.

Non-Linear DSGE Model Solving

• Projection Methods: Solve non-linear DSGE models using advanced numerical techniques:
  • Galerkin Method: Project solutions onto a basis of functions for accurate approximation.
  • Collocation Method: Solve at specific points to approximate the policy function.
  • Least Squares Method: Minimize residuals to find optimal solutions.

Getting Started

To use Econometron, install it via pip:

pip install econometron

Example: Fitting a VAR Model

from econometron.Models.VectorAutoReg import VAR

# Load your time series data
data = ...  # Your multivariate time series data
model = VAR(data=data,max_p=2,check_stationnarity=True)
results = model.fit()

For detailed documentation, tutorials, and examples, visit the Econometron Documentation.

Why Econometron?

• Comprehensive: Covers a wide range of econometric models, from classical VAR to cutting-edge neural network approaches.
• Flexible: Supports both statistical and machine learning-based methods for time series analysis.
• Robust: Implements state-of-the-art estimation and identification techniques for reliable results.
• User-Friendly: Designed with Python's ecosystem in mind, integrating seamlessly with libraries like NumPy, Pandas, and PyTorch.

Code of Conduct

We are committed to fostering a welcoming and inclusive community. All participants are expected to:

• Be respectful and considerate in interactions.
• Avoid harassment or discriminatory behavior.
• Use constructive feedback and maintain professionalism.
• Respect the community and project’s guidelines.

Violations may result in removal from the project or community channels. Please read the full Code of Conduct for details.

Contributing

Contributions are highly valued. To contribute:

1. Fork the repository.
2. Create a new branch for your feature or bug fix.
3. Implement your changes with clear, well-documented code.
4. Run all tests to ensure stability.
5. Submit a pull request describing your changes in detail.

For more information, see the full Contributing Guide.

License

Econometron is licensed under the MIT License. See the LICENSE file for more information.