Overview

MLaC (Machine Learning as a Code) is a python code corresponding to each ML pipeline stage

User can either directly use that code for model training or can create a docker container and deploy/execute using docker compose, kubernet or kubeflow. This components code is generated using the AION. Users will have to train the model in AION and based upon the configuration, the code will be generated which can further be used to retrain the model in production environment. Based upon the configurations (Basic and Advance), automatic code will be generated to provide the facility for running the code as a pipeline, on the local machine or on the cloud.

Pipeline Components

• DataIngestion: Read the data from specified location and save it for the later use.
• DataTransformation: Data cleaning and transformation is performed.
• FeatureEngineering: Feature reduction or feature selection techniques applied.
• ModelTraining: Train models based on the algorithms selected and Feature Engineering options and create separate folder for each of the model to run them in the parallel.
• Prediction: Best model is selected based on the score and perform prediction on the provided data.

Supported Platforms

• Microsoft Windows
• Red Hat Linux

Prerequisites

For generating the Container, Docker application should be installed on your system.