Lab 1 for ID2223 Course @ KTH

Keywords: scalable machine learning, data engineering, classification, model ensemble

Team

Name: GoodOnions
Components: Federico Bono, Daniele Cipollone

Project description

This laboratory is composed by two main tasks: a guided implementation of a ML pipeline for a classification task for the Iris dataset and an autonomous design and implementation of a ML pipeline for either regression or classification task for the Wine Quality dataset. In details we will need to build two small applications, one for on-demand inference and the other to monitor the model performance over time. We will also need to develop a script that generates a new entry every day to test our system.

Goal

The main goal is to design and implement the ML pipeline for classification on the Wine Quality dataset, we have followed this steps:

1. Exploratory Data Analysis (EDA)
2. Feature engineering and selection with data cleaning
3. Backfill training data
4. Setup of the training pipeline
5. Setup of the inference pipeline
6. Setup automatic daily entry generator
7. Setup of the two frontend applications
   1. Realtime inference
   2. Model performance monitor

Public links

1. Realtime inference
2. Model performance monitor

References

• Related Kaggle challenge
• Complete analysis by Marcelo Marques