QMIB labs — MATH60033A

One file per lab. Open it, work through it, submit. Python runs inside the page — numpy, pandas, statsmodels, scikit-learn, matplotlib — so there is nothing to install and nothing to download except this page. Each lab times itself, checks your code as you go, and at the end sends your answers, your code, your figures and your timings to the gradebook.

Labs follow the session's own practice deck: reproduce a result, apply the method, break an assumption on purpose, then write what it does and does not support. That last part is the one that carries the marks.

These are optional. They are the at-home knowledge check described in §5 of the syllabus: a private way to find out whether you followed the session, and a route to participation marks that does not require speaking in a room of thirty. They are never a prerequisite for the next lecture.

Session 02 Exploratory data analysis: fundamentals Which summary of your key variable would you defend in print? 65 min8 code cells core Session 03 Regression: Adequacy, Validity, and Robustness Which model would survive a referee? 69 min7 code cells core Session 04 Logistic Regression: Binary, Ordinal, and Multinomial Can we predict a discrete outcome honestly? 66 min8 code cells loans, redwines, ologit, hsbdemo Session 05 Regularisation: ridge, lasso, elastic net Of two hundred indicators, which few actually carry the signal? 67 min9 code cells fredmd Session 06 Regression: Advanced Considerations Does your finding survive the structure of your data? 66 min7 code cells panel, realestate, glassdoor Session 07 Principal Component and Factor Analyses How many distinct dimensions does your angle really have? 65 min6 code cells movies, efa Session 08 K-Nearest Neighbours and the Bias-Variance Trade-off Does flexibility buy you anything on your own data? 65 min6 code cells core Session 09 Structural Equation Modelling What is the construct behind your indicators? 60 min5 code cells efa Session 10 Causal Inference I: Counterfactuals, Randomisation, Matching Can your project support a causal claim at all? 61 min7 code cells core Session 11 Causal Inference II: Difference-in-Differences What is your counterfactual, and would anyone believe it? 58 min6 code cells core

If the wifi is unreliable

The first time any lab runs it fetches the Python runtime once, about 12 MB, and the browser caches it — every later lab starts instantly. If you were given a folder with a pyodide/ directory beside these files, nothing is fetched at all and the labs work with the network off. Submitting still needs a connection; if there is none, the lab downloads your submission as a file and tells you so.

What is recorded

Your name, your HEC email and matricule, what you typed, what your code printed, the figures it drew, and how long each section took. Nothing else. The clock pauses by itself after four minutes without activity, so it measures work rather than elapsed time.