Day 2: Lightning Talks
summary
Grab concrete takeaways: an invitation to submit to Budapest PyData and a community project connecting activists and technologists; a Python time-series library with a one-liner API that composes regression-based trend and seasonality components; Japanese NLP tips using MeCab and SentencePiece; a pip-inspection demo and a proposal for an unsafe fixed-width reader that claims a 600% speed-up and 10× memory reduction by reading into numpy and avoiding decodes; an LSTM beer-recipe generator that exists but doesn’t make great beer yet.
AI-written summary · about
chapters (10)
- 0:00Budapest BI Conference welcome
- 1:53Radical Engineers for change
- 4:43Python time series library
- 7:39Japanese for data work
- 13:17Smartphone dependence and inequality
- 17:09Python package dependencies
- 21:51Data science skills report
- 26:04Optimizing Pandas for fixed-width files
- 33:22AI over-engineering beer recipes
- 38:02Closing notes and feedback
