Data: Faithful or Traitor?
summary
BBC uses 150 billion rows a month of Piano analytics and 34 million signed‑in accounts alongside Chartbeat, BARB panels and national surveys; a random forest predicts postcode‑level school holidays and validates against weather disruptions; inverse propensity weighting estimates campaign lift without A/B tests; Kaleidoscope maps demand spaces from BARB collaborative filtering; the Wall forecasts slot‑level viewing from schedule features; Python generates fast personalized pre‑rolls for archived titles. Some models perform well, others remain open questions for audience research.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 2:14BBC mission vs analytics
- 4:05BBC data science projects
- 10:02School holiday viewing model
- 13:13News genres clustering
- 15:20News sentiment analysis
- 18:14Demand spaces for news
- 20:24Kaleidoscope TV demand
- 22:29The Wall forecasting tool
- 24:39Marketing effectiveness with propensity
- 28:15Personalized marketing trailers
- 31:43BBC Bitesize quiz during COVID
- 34:12Summary
related talks
52mAB-tests, lies, damned lies, and statistics.
01h 29mShow, Don’t Tell How Data Is Changing Hollywood
52mMaking Better Decisions with Data
53mDemocratising Attention Data at theguardian.com
10mHigh Order Bit
11mThe Value of Social (for) TV
39mEmphasising Relationships in the BBC's Data Using Technologies of the Semantic Web
27mHow to be data-driven when you aren’t Netflix (or even if you are)