Project Overview

The goal of this project was to build an ML model that predicts the delivery time of a product purchased at a given time. The project included the following steps:

  1. Determining business objectives with the client.
  2. Data cleaning, EDA, data visualization, feature selection, feature engineering, etc.
  3. Preparing model for deployment as a web service.

Example POST request

MLOps Implementation

Implemented MLOps practices to streamline the delivery of the model:

  • Organized the codebase into a readable structure.
  • Used Prefect for monitoring data processing and training pipelines.
  • Code reruns only modified components of the machine learning pipeline.
  • Trained model can be easily deployed as a web service in Flask.

Training pipeline monitoring using Prefect

Resources