enhancesa 0.1a0

Python micro-package for enhanced statistical analysis

enhancesa

Enhancesa is a collection of tools for a better and more simplified statistical analysis in Python. It primarily aids in manual analysis and prediction tasks that use packages like Statsmodels and Scikit-learn in their workflow.

For example, Enhancesa provides answers to questions like: Which subset of features gives me the lowest error rate in an ordinary least squares model? What are estimates of population mean and standard deviation using bootstrap resampling? And etc.

Upcoming features

• Partial least squares (PLS) regression
• Principal components regression (PCR)
• Subset selection plots
• Additional test statistics in bootstrap resampling

Motivation

Enhancesa is a result of solutions to exercises in the book Introduction to Statistical Learning by the Tibshirani et al. When going through the exercises, I found Python, unlike R, lacking in providing convenient functionalities. At this stage, this package is simply a collection of functions I used in my solutions to exercises in the book.

Installation

Enhancesa can be installed from the PyPI package repository.

$ pip install enhancesa

Quick glimpse

>>> import numpy as np
>>> import enhancesa as esa
>>> # Create some dummy data
>>> x = np.random.normal(size=100)
>>> # Compute test statistics with bootstrap resampling
>>> esa.bootstrap(x, iters=1000)
Estimated mean: -0.025309
Estimated SE: 0.095531
dtype: float64

Find out more about the full set of features in the documentation.

Issues & improvements

• Possible to further reduce dependencies.
• boostrap method can be improved by adding estimates of more test statistics of interest.
• Use Poetry for package and dependency management, which uses pyproject.toml recommended by PEP 518.
• enhancesa.SubsetSelect will give NotImplemented error if X input is a Numpy array.

License

This package is licensed under an MIT license.