This app allows users to apply a variety of AI/ML tools without the need for coding. It integrates multiple frameworks designed specifically for AI/ML workflows, streamlining complex processes into a user-friendly interface. The app also includes predefined functions for data management, enabling users to easily handle data preparation, cleaning, and transformation tasks. Additionally, it offers robust visualization capabilities, allowing users to explore and interpret results effectively, all within a seamless, integrated environment.
The app is designed to provide end-to-end AI/ML workflow. Specifically:
- Upload
- Exploration
- Transformation
- Merging
- Visualization
- Transformation
- Data partitioning
- Feature engineering
- Model selection
- Hyperparameter tuning
- Model comparison
- Prediction
- Deployment
The APP depends on the R package Rautoml which handles most of the back-end R functionalities. To start the app, you would first need to have the package installed.
devtools::install_github("aphrc-nocode/Rautoml")
Clone the repo first :)
Prepare your Python environment (FastAPI + PyCaret)
Step 1: Set up a new virtual environment
- Create a new virtual environment
python -m venv env1
- Load virtual environment created on Windows
env1\Scripts\activate
- Load virtual environment created on Linux
source env1/bin/activate
Step 2: Install libraries needed
pip install fastapi uvicorn pycaret pandas python-multipart shap
The APP depends on some Python libraries :
- FastAPI : Create API easily and fastly
- Pycaret : Module for ML no-code
- Uvicorn : It help to launch FastAPI
- Python-multipart
To start the app, you would first need to have the package installed.
main.py file contains code to create an API with FastAPI and Pycaret. Go to command line, move to py folder and run this following command to Launch FastAPI.
uvicorn main:app --reload
Open either server.R or ui.R from R Studio and then R the app.
From the terminal, run:
make runapp
Our workflow uses make pipelines to sync various components of the app. But you can also use standard Git commands or GUIs. See Makefile.
- To get started, run:
make Makefile
- To update (pull or push), run the command below. This will show a pop-up to update the commit message or use the default (discouraged :()
make sync
- Steve Cygu
- Michael Ochola
- John Lunalo
- Steve Cygu