Exploring StackOverflow Data

on YouTube ↗47m

" summary

Stack Overflow’s quarterly dump (≈135 GB, ~5M users, ~1M with fillable locations) gets parsed with type providers (CSV, JSON, HTML) to infer schema and call Bing for geocoding; weekend-index shows languages like F at ~0.6 and leisure technologies (Minecraft, Pygame) >1. Country concentrations give Denmark ≈82 F programmers/million and Iceland ≈185 R programmers/million, with Dominican Republic a sampling outlier. High-dimensional tag relationships use t-SNE and produce an interactive F→JS visualization via Fable. Word embeddings via word2vec reveal language analogies; regression explains ~40% of answer score and one high-leverage user skews results.

AI-written summary · about

" chapters (11)

1. 0:00Start
2. 0:16Disclaimer
3. 3:57How Standard Post on Stack Overflow Looks
4. 4:21Which Tags Are Most Common on Stack Overflow
5. 8:11When Do People Ask or Answer Questions on Stack Overflow
6. 21:00Type Providers
7. 23:21Where Are the Most Programmers per Million for each Language
8. 24:45Huge Sampling Bias
9. 25:18Sampling Biases
10. 27:55Visualization
11. 39:31Stack Overflow and Meritocracy

" related talks

• 31mA Recent History of Python Through Stack Overflow Questions and AnswersSam Hames · PyCon AU 2023
• 36mThe mysterious correlation: a detective storyEvelina Gabasova · Codemotion Milan 2017
• 43mBehind the Scenes at Gapminder: A Story about Data and VisualizationsJasper Heeffer · Øredev 2018
• 35mExposing AlgorithmsJennifer Stark · PyData DC 2016
• 26mBy the Numbers: Python Community Trends in 2017/2018Dmitry Filippov, Ewa Jodlowska · PyCon 2018
• 14mInformal Learning Communities: The Other Massive Open Online 'C'Michael Lynch, David A. Joyner, Will Hudgins, Ash Schmal, Harsh Sikka, Michael Swenson · Learning at Scale 2020
• 20mThe next million R usersCarl Howe · rstudio::conf 2019
• 25mUX, DX, DSX: Developers and Data Scientists as UsersDot Gong · UXDX Europe 2019