The mysterious correlation: a detective story
summary
Stack Overflow survey shows spaces associated with higher median salary but deeper analysis using linear models finds linear regression and adjusted R^2 equal with and without indentation, so tabs/spaces add no predictive power. Exploration reveals a bimodal salary distribution among space users and that version control explains much: git vs TFS correlates with higher pay. Investigation uses regression, coefficient comparison, density plots and dataset cleaning (fixing monthly vs annual entries) and the pattern points to company/language/ecosystem, not causation.
AI-written summary · about
chapters (6)
- 0:00Start
- 2:49CORRELATION AND CAUSATION
- 4:52CAUSAL LANGUAGE FOR CORRELATION DATA
- 13:43COLLINEARITY?
- 16:56POTENTIAL EXPLANATION?
- 23:40WHY IS VERSION CONTROL SO IMPORTANT?
related talks
47mExploring StackOverflow Data
22mUnicoRns are real
48mHuman data interactions: experiments in how humans analyze data
22mR & Space : GI Science with R
26mSridhar Telidevara: Professor, Great Lakes Institute of Management at Cypher 2017 - Part 1/2
15mThe unexamined bytes are not worth coding
44mAutomating Analysis
01h 03mWhat Can We Learn From the Results of C++ Community Surveys?