Gomat Markup Optimization

Gomat Markup Optimization is a system developed at GoMaterials to recommend optimal selling markups for e-commerce transactions. It trains conversion probability and markup optimisation models to maximise revenue and conversion rates.

Features

• Data Preparation & Feature Engineering: Balances datasets, handles class imbalance and generates synthetic data.
• Modelling: Uses CatBoost and Random Forest algorithms to train conversion probability and markup space optimiser models.
• API: Provides a FastAPI inference API to deliver recommendations.
• Experiment Tracking: Integrates MLflow to track experiments and manage models.
• CI/CD & Containerization: Implements continuous integration and deployment pipelines with Bitbucket and Azure DevOps; containerises the application with Docker.
• Deployment: Deploys the solution to Azure Machine Learning Studio.

Usage

The project contains modules for data preprocessing, model training, hyperparameter tuning and inference. You can adapt these components to your own dataset and environment. Use the FastAPI endpoints to obtain markup recommendations for your transactions.

Results

By recommending optimal markups, the system increased conversion rates and revenue for GoMaterials’ e-commerce platform.