DataLab Project
The DataLab Project is the applied component of the course. You will use the concepts introduced in the self-study material to solve a real machine-learning problem.
The project follows an iterative CRISP-DM workflow. Rather than completing each part of the machine-learning process only once, you will revisit data understanding, data preparation, modeling, and evaluation as your understanding of the problem develops.
As you progress through the project, the level of guidance gradually decreases.
At a glance
| Data | Guidance | You end with | |
|---|---|---|---|
| Iteration 1 | starter_dataset.xlsx: 80 molecules, ready for modelling |
Step by step | Your first classification, regression and clustering models |
| Iteration 2 | The raw HOPV files: 350 molecules | Structured notebooks; more decisions are yours | Your own dataset and improved models |
| Iteration 3+ | HOPV + a lab database + the Kaggle molecules | Pointers only | A Kaggle score, and a model you can explain |
There are no deadlines and no grades: work at your own pace. The Kaggle leaderboard tells you how well your solution works on molecules it has never seen. When a notebook mentions what the client expects, such as a target score or a justification of your choices, treat it as advice rather than a requirement.
Self-study and DataLab go side by side. A good rhythm is to finish Self-study theme 1 before Iteration 1, themes 2 and 3 before Iteration 2, and to use themes 4 to 6 as your toolbox in Iteration 3+. The 8-week plan puts this into a schedule.
Iterations
Iteration 1 — First ML Cycle
Go through the complete machine-learning workflow for the first time and build your first classification, regression, and clustering models.
This iteration is highly guided. The goal is to experience the complete workflow quickly and establish your first working models rather than optimize performance.
Iteration 2 — Go Deeper
Revisit the machine-learning workflow in greater depth. Develop a stronger understanding of the problem and data, improve your data preparation, and build more considered classification, regression, and clustering models.
This iteration is still structured, but you will begin making more decisions yourself and applying concepts introduced in the self-study material.
Iteration 3+ — Compete, Iterate, Improve
From this point onward, the project becomes open-ended.
Use your existing models as a starting point and compete on the Kaggle challenges. Analyze your results, identify weaknesses, form hypotheses, and decide what to improve next.
There is no fixed sequence. You may revisit any part of the machine-learning workflow, including data understanding, data preparation, feature engineering, validation, model selection, and hyperparameter tuning.
Continue iterating based on evidence until you are satisfied with your solution.