@jonathan.interviews: Bayesian updating is one of the most tested statistics concepts in data science interviews — and many candidates get it wrong. New series: we practice interview questions together so you can train for the stats knowledge module at top tech companies. Here’s a real one framed around Uber. Given a 10% chance a user rides during morning rush, 20% during evening rush, and 50% of an evening ride if they had a morning ride earlier that day — what’s the probability someone took a morning ride if we observed they took an evening ride? The answer is 25%. You flip the conditional using Bayes’ theorem: p(morning) times p(evening|morning), divided by p(evening). That’s 10% times 50% divided by 20%. The trap is assuming p(evening|morning) and p(morning|evening) are interchangeable. They’re not. #datascience #datascienceinterview #statistics #techinterview #dataanalytics

Data Science Interviews

Open In TikTok:

Region: TH

Friday 20 March 2026 14:26:46 GMT