A Cowork plugin that generates fresh Python data science practice problems on demand — grounded in real reference books, set in real-world scenarios, and structured to build skills progressively.
Ask Claude for practice problems on any data science topic and it will generate a self-contained practice file (.py or .ipynb) with:
- 1 worked example — complete solution with explanation
- 10 problems — ★☆☆☆☆ to ★★★★★, two per difficulty tier
- 1 challenge problem — combines multiple concepts
- A test harness — run the file to see
[PASS ✓],[FAIL ✗], or[TODO ]for each problem
Problem structures vary by difficulty: fill-in-the-blank at the low end, word problems and open-ended multi-step analysis at the high end. Hints cite specific book chapters rather than just naming functions.
Give me practice problems on hypothesis testingI want to practice pandas groupby and aggregationMake me some numpy broadcasting exercisesPractice problems on logistic regressionGive me some Python practice problems(picks a topic for you)
Double-click python-practice.plugin to install in Cowork.
The skill works out of the box without any additional setup — it will generate problems and cite relevant chapters from memory.
For the best experience, clone the reference books it draws from into a reference/ folder in your Cowork workspace. When the books are present, the skill reads the actual chapters before writing problems, which produces more pedagogically grounded exercises.
# In your katas/ or practice/ folder:
mkdir reference && cd reference
# Python for Data Analysis (Wes McKinney)
git clone https://github.com/wesm/pydata-book
# Python Data Science Handbook (Jake VanderPlas)
git clone https://github.com/jakevdp/PythonDataScienceHandbook
# ThinkStats: Exploratory Data Analysis (Allen Downey)
git clone https://github.com/AllenDowney/ThinkStats
# Data Science from Scratch (Joel Grus)
git clone https://github.com/joelgrus/data-science-from-scratch
# Bayesian Methods for Hackers (Cameron Davidson-Pilon)
git clone https://github.com/CamDavidsonPilon/Probabilistic-Programming-and-Bayesian-Methods-for-Hackers
# Efficient Python Tricks and Tools for Data Scientists (Khuyen Tran)
git clone https://github.com/khuyentran1401/Efficient_Python_tricks_and_tools_for_data_scientistsAll six books are freely available and open-source.
The skill draws on all six books and covers:
| Area | Topics |
|---|---|
| Python fundamentals | Builtins, itertools, collections, comprehensions |
| NumPy | Arrays, broadcasting, ufuncs, linear algebra |
| pandas | Indexing, cleaning, strings, time series, groupby, merge/reshape |
| Visualization | matplotlib, seaborn |
| Statistics | Descriptive stats, distributions, correlation, regression, hypothesis testing, confidence intervals, estimation |
| Machine learning | scikit-learn workflow, preprocessing, kNN, Naive Bayes, linear/logistic regression, decision trees, random forests, PCA, clustering |
| Bayesian methods | PyMC, MCMC, priors, loss functions |
| Other | NLP, network analysis, gradient descent, recommender systems |
MIT