Dropping In Surf Show Podcast Data Science in Surfing with Daphne Teh
Sep 10, 2024
Daphne Teh, a data scientist and Bocconi professor who applies AI to business problems, explores surfing use cases. She talks about data quality, model choices, YouTube analytics for surf channels, timing and location effects, personalized surf forecasts and recommender-style reports, wearables and sensor data, and combining multi-source data to boost performance and sponsorship value.
AI Snips
Chapters
Transcript
Episode notes
Collect Widely Then Model Narrowly
- When possible, collect many low-cost variables up front because you may need them later.
- Then be selective when modeling: include only variables relevant to the question to avoid garbage-in garbage-out.
YouTube Benchmarking For Surf Channels
- Daphne scraped ~41 YouTube surf channels to benchmark content and posted frequency, then compared channels like Kale Brock for channel health.
- She used view averages and standard deviation to spot consistent high performers versus one-hit peaks.
View Variation Reveals Channel Health
- Variation (standard deviation) in views matters as much as average views to judge channel health.
- A good channel has high average views with low variation rather than one viral hit and many low performers.