~/projects/coursework $ cat README.md
Coursework & playgrounds
Smaller things from my first semester at UCPH.
- Python
- Jupyter
- NLTK
- NumPy
- Matplotlib
- scikit-learn
Multilingual text processing (LP1)
For Language Processing 1, I compared literary texts in English, Swedish and Ukrainian: my own regex tokenizer against NLTK, then POS tagging, stemming versus lemmatisation, and word frequencies. What surprised me most was how much information gets lost when you normalise text.
Scientific programming from first principles
For Scientific Programming, I implemented maths from scratch in Python: factorials, approximations of constants and square roots, matrix multiplication and recursive determinants. For each one, I compared my version with an AI-generated one, looking at readability and input checks.
Sentiment-classifier playground
Outside the course, I used Codex to extend a course notebook with a simple sentiment classifier: probes with invented sentences, error inspection and word-by-word explanations. It shows pretty quickly that a bag-of-words model can't tell “not good, bad” from “not bad, good”.