Assessment
Book signatures
I analysed Project Gutenberg texts using metadata, letter counts and word-length features, with the slightly pleasing idea that books have statistical fingerprints.
OCOM5100M / Programming for Data Science
Text at Scale
My final assessment built a Python data-analysis pipeline around Project Gutenberg: bulk text ingestion, metadata joins, derived letter and word-length statistics, and visual analysis of a corpus large enough to force practical thinking.
Data work turns into systems work almost immediately: file formats, storage, repeatability and provenance are what decide whether anyone should trust the analysis.
Tools
Python as workshop kit
The module covered notebooks, pandas, NumPy, plotting, data loading, cleaning and producing reproducible analysis rather than one-off graph magic.
Lesson
Data is never “just there”.
The bulk corpus needed careful handling: supported downloads, derived files, joins, storage costs and explicit assumptions about what the metadata did and did not contain.
Carry-forward
Data work becomes systems work
This became useful later in robotics and deep learning work, where dataset shape, provenance and derived features mattered as much as the model.
