Practical day 1
Hands-on workflows with NLTK, scikit-learn, and Gensim — POS tagging, TF-IDF, and LDA topic models — plus guidance on the assignments
Open slidesOverview
A working session, not a lecture: laptops open from the start. Three short demos put the lecture 1–3 toolkit to work — part-of-speech tagging with NLTK, TF-IDF with scikit-learn, and a full LDA topic model with Gensim on 2,000 newsgroup posts (the complete pipeline: clean → tokenize → filter → dictionary → bag-of-words → model). The second half is a supervised work block on assignment 1 with your group; we circulate.
Bring
- A laptop with a working Python setup — see Tools; if your environment is broken, today is the day we fix it
- Your group’s assignment problem — data, questions, half-broken code all welcome
Materials
- Slides: use the Open slides button above
- Notebook: practical_day_1.ipynb —
the three demos plus working-with-code tips. The LDA demo fetches the
20 Newsgroups dataset automatically; no data download needed. First run
requires the commented-out
nltk.download(...)lines.
Before lecture 6
- Read Kozlowski et al. (2019) — discussed in the embeddings lecture
- Keep your notebook environment alive — practical day 2 builds on it