Colin Morris writes about data and language experiments, including lyric visualizations, Wikipedia oddities, and neural networks that generate text and names.
HN commenters link it for its thorough, funny writeups on generative models and for the least-viewed Wikipedia article hunt.
Fun to see the Aubrey-Maturin series mentioned here; that was indeed the first thing I thought of.
Killick's use of "which" always struck me as odd. I'm not entirely sure if it's the same grammatic construct as the one in the article. Here [1] is a discussion about it. The top answer argues it's just a meaningless discourse marker. It was also covered in Mark Liberman's Language Log [2].
Apparently the conjunctive "which" is increasing in usage as a discourse marker [3].
[1] https://english.stackexchange.com/questions/245363/what-is-the-name-of-the-grammatical-device-where-18th-and-19th-century-sentences
[2] http://itre.cis.upenn.edu/~myl/languagelog/archives/000118.html
[3] http://colinmorris.github.io/blog/which-discourse-marker
If you're looking to do more analysis on this data, it would be cool to see a correlation matrix for different articles of clothing in different categories. E.g. looking at how frequently you wear certain tops/bottoms together, or bottoms/shoes.
https://www.displayr.com/what-is-a-correlation-matrix/
Another example of this chart I saw recently was examining compound curse words on reddit:
https://colinmorris.github.io/blog/compound-curse-words#the-matrix-of-pejoration
What you want is a language model. This will give you the probability on a word by word basis.
Something like [1] is pretty much state-of-the-art. It's worth noting that the kind of writing you are doing change the probability significantly. [2] shows this quite well.
[1] https://colinmorris.github.io/lm-sentences/#/billion_words
[2] https://colinmorris.github.io/lm-sentences/#/brown_romance
I'm not sure how easy it is to grok these by playing with them. If you're confused, the "About" section has some explanation of how the matrices are constructed and how to interpret them: https://colinmorris.github.io/SongSim/#/about
Also, in case it's not clear, you can press the "+" button next to the song dropdown to open a textarea where you can paste arbitrary lyrics.
Reminded me of least viewed pages on wikipedia collections or neglected articles...
See:
In search of the least viewed article on Wikipedia (2022)
https://colinmorris.github.io/blog/unpopular-wiki-articles
(Some discussions: https://news.ycombinator.com/item?id=31524943, https://news.ycombinator.com/item?id=37955600)
This is a pretty thorough description of generation, not just the classification problem that is the usual introduction to machine learning.
I just loved the map of the generated place names [1] the illustrations are hilarious.
[1] http://colinmorris.github.io/assets/rbm/kicksville.png