regpy

A small library for creating and plotting regression models.

regpy aims to make it as painless as possible to create and plot regression models, making it a great tool for incredibly rapid prototyping over small datasets or as an educational tool for demonstrations.

Setup

Run pip install -e . in the root directory and you should be good to go!

Getting Started

To begin using regpy, import the package:

import regpy.linreg as lrp

To ensure that the the module as been correctly imported, run it. If you get a ModuleNotFoundError, refer to the Notes section of this README.

To create a linear regression, initialize a LinearRegression object, which requires two lists. The only restriction on these lists is that they cannot be empty.

import regpy.linreg as lrp
# This should create a linear regression that can be represented as y = x on a graph
linReg = lrp.LinearRegression([1, 2, 3], [1, 2, 3])

To make a prediction based on the regression line, use the predict() method, which is a member of the LinearRegression class.

import regpy.linreg as lrp
linReg = lrp.LinearRegression([1, 2, 3], [1, 2, 3])
print(linReg.predict(-1)) # Will print -1.0
print(linReg.predict(0)) # Will print 0.0
print(linReg.predict(10000)) # Will print 10000.0

You can also access the R-Squared value for the regression using the public fields on a LinearRegression object.

import regpy.linreg as lrp
linReg = lrp.LinearRegression([1, 2, 3], [1, 2, 3])
print(f"R^2 = {linReg.rSquare}") # R^2 = 1.0

Finally, printing a LinearRegression will give you the line equation in the form of y = {beta}x + {inty}, both rounded to two decimal places (beta and inty, as well as the domain and range of the dataset, are also freely accessible).

import regpy.linreg as lrp
linReg = lrp.LinearRegression([1, 2, 3], [1, 2, 3])
print(linReg) # y = 1.00x + 0.00

Plotting

Plotting a regression line is done internally using matplotlib, and only requires two functions to be called.

import regpy.linreg as lrp
linReg = lrp.LinearRegression([1, 2, 3], [1, 2, 3])
linReg2 = lrp.LinearRegression([2, 3, 4], [2, 3, 4])
lrp.plotLinReg(linReg) # queues the first regression line to be plotted
lrp.plotLinReg(linReg2) # queues the second regression line to be plotted
lrp.drawPlots() # displays the plot

Plotting the scattered points of the dataset alongside the line is optional, and so is setting a label for the x and y axes. By default, the scatter is always plotted, and the x and y axes are labeled "x-values" and "y-values" respectively. The title of each plot will always be the string representation of the given LinearRegression.

import regpy.linreg as lrp
linReg = lrp.LinearRegression([1, 2, 3], [1, 2, 3])
lrp.plotLinReg(linReg, xLabel="x-axis", yLabel="y-axis", scatter=False) # optional parameters
lrp.drawPlots()

Notes

Some may have still have issues importing this package after installation, to which I would recommend using a virtual environment:

Install venv using pip install venv or pip3 install venv

To create a virtual environment, run python3 -m venv venv

Activate it with venv/Scripts/activate (for Windows) or source venv/bin/activate (for Linux-based)

Run pip install -e . again in the root directory and it should work then. To close the virtual environment just use deactivate

regpy provides a set of tests in the test/ directory to ensure that it doesn't just run on my machine. If all tests pass you can be reasonably confident it will work for your use cases

This library is written purely in python, so expect performance bottlenecks should you decide to apply this library to large-scale data sets. It is also a WIP, so there may be new types of regressions that are added to the library in the future.