STATINF

This library aims at re-implementing standard statistical tools (such as OLS, logistic regression, Neural Network) and is built on top of numpy for handling data and jax for differential computing. The objective is to implement new methodologies from research projects on these models. The library also provides a data generator for linear and binary data.

The library is pip-installable and the source code is available on my Git. For any question or suggestion of improvement, please contact me.

Installation

You can get STATINF from PyPI with:

pip install statinf

The library is supported on Windows, Linux and MacOS.

STATINF tries to use the least number of dependencies possible:

pandas: used to convert data frames into arrays.
numpy : main library for data handling and matrix operations.
scipy: probability density functions.
jax: matrix operations and back-propagation for Deep Learning models.
matplotlib: plots of training performances.
pycof: basic information printing.

Once your dependencies are installed, you may need additional steps to enable GPUs (if eligible), see FAQ.

Indices and tables

Modules available

Distributions

Discrete probability distributions
- Discrete distributions
- Example

Econometrics

Linear regression
- OLS
- Examples
  - OLS
  - LinearBayes
Generalized Linear Models
- GLM
- Example

Statistics

Machine Learning

Model performance
- Binary accuracy
- Example