AutoML - Application

The project

The project is an AutoML solution to train Machine Learning models with little human interference. This is a extension of the App with new features.

Automated machine learning, also known as AutoML, is the process of automating the time-consuming and iterative tasks of machine learning model development. With it, data scientists, analysts, and developers can build ML models with high scale, efficiency, and productivity while supporting model quality.

Technologies

• Pycaret (lib for AutoML)
• Pandas
• Streamlit
• Docker
• App Engine (Google Cloud Platform) for Cloud Deployment

Features

• Load data for classification problems
• Analysis of dataset
• Select predictor variabels
• Train model
• Get the best model

How to run

Run in a virtualenv

git clone https://github.com/ozseniorcl13/AutoMLApp
cd AutoMLApp
python -m venv venv
source venv/bin/activate  (for linux)
venv/Scripts/activate     (for windows)
pip install -r requirements.txt
streamlit run app.py

Run a container

git clone https://github.com/ozseniorcl13/AutoMLApp
cd AutoMLApp
docker build -t automlapp .
docker run -p 5000:8080 automlapp
Access in your browser: http://localhost:5000

To do

• Select the type of analyse (minimal ou complete)
• Select setup configuration for training
• Add support for regression problems
• Export the results of training