The Project

In the project, you are expected to train a Machine Learning model on a dataset of your choice, evaluate it and report the results.

Steps

Do the following:

  1. Form a team of 2-4 people

  2. Find a dataset suitable for Supervised Learning, e.g. on Kaggle

  3. Explore the data (view, create a few plots)

  4. Train a model

  5. Evalute the quality of the model

  6. Report your results (the submission should include the code and a report of about two pages)

Note

The dataset should contain at least 1000 data points. That excludes the following datasets: Titanic, penguins, Boston Housing and everything that was used in class.

Hint

Feel free to use any programming language you like.

Assessment Criteria

I will check each of the following:

  • Is the goal of the project clearly described?

  • Were basic data quality checks done (for size, missing data, distribution of columns etc.)?

  • Were three plots created describing the data?

  • Was a suitable Feature Engineering method used?

  • Was a model trained on the data?

  • Was another model or baseline model trained as well?

  • Was a main quality metric calculated?

  • Was a deeper analysis of model quality done using other metrics or plots?

  • Was the model checked for overfitting!?

  • Was cross-validation used?

  • Is the code in a GitHub repository?

  • Is the code clean and tidy?

  • Is the report well-written and concise?

  • Is there anything that is particularly excellent? (up to 2 points)

Warning

Do not copy-paste solutions done by other people. I am using a plagiarism checker.

If you let yourself inspire from code written by other people, you must mention these sources in your report. Do it!

While ChatGPT and other LLMs might help in steps 3 and 4, using one in the other steps is a bad idea for studying and outright dangerous in a real-world project. Don’t!