sharozhaseeb / RegressorHub

This is an easy scratch implementation for linear and logistic regression

RegressorHub

This repository contains IPython notebooks that demonstrate linear regression and logistic regression models implemented from scratch in Python.

Contents

• main.ipynb: Implements linear regression and logistic regression to predict student performance scores. Covers data preprocessing, model training, evaluation, saving and loading.

• automl_main.ipynb: Implements auto-sklearn to perform model training and inference.

Usage

The notebooks contain code and explanations for each step. They can be run end-to-end to train models, evaluate them, and save the trained models.

The trained models are saved as Pickle files that can be loaded for inference later.

Requirements

The notebooks require the following Python packages:

• numpy
• pandas
• scikit-learn
• matplotlib
• pickle

References

The implementation is based on concepts from:

• Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow by Aurelien Geron
• Introduction to Machine Learning with Python by Andreas C. Muller and Sarah Guido

Next Steps

• Normalize data(Z-Score, MinMax)
• Hyperparameter Tuning;
• Adding WandB Logs;
• Maybe a new learning rate?;
• Push to hub
• save locally