Machine Learning Regression models

Demo app

Launch the app

App info

This repository creates a Data Web App that enables users to interact with several hyper-parameters of Support Vector Regression algorithm from Sci-kit learn Python package. The created ML model is evaluated based on two performance metrics, namely: Coefficient of determination and mean squared error. The hyper-parameters are:

1. kernel
2. Regularization parameter C
3. Epsilon
4. Degree of polynomial kernel
5. Kernel coefficient gamma
6. Independent term
7. Hard limit on iterations
8. shrinking heuristic
9. size of the kernel cache
10. verbose output
11. Tolerance

Default Dataset

Regarding the default dataset, the concrete slump test measures the consistency of fresh concrete before it sets. It is performed to check the workability of freshly made concrete, and therefore the ease with which concrete flows. It can also be used as an indicator of an improperly mixed batch.

Reproducing the App

1. First, we create a virtual Python environment called my_venv

  python3 -m venv my_venv

2. Then, we activate the virtual environment

source path_to_your_virtual_environment/bin/activate

3. After getting to the virtual environment's file, install prerequisite packages

wget https://raw.githubusercontent.com/GeorgiosDolias/ML_app/main/requirements.txt

and

pip install -r requirements.txt

4. Dowload and unzip contents from Github repo

Dowload and unzip contents from https://github.com/GeorgiosDolias/ML_app/archive/main.zip

5. Launch the app

streamlit run ml-app.py

Requirements

Package	Version
streamlit	0.88.0
pandas	1.1.3
sci-kit learn	0.23.2

Useful Resources

1. Youtube tutorial from Chanin Nantasenamat (Data Professor)
2. Details of default dataset