IAB Canada Roadshow 2015: Turning Machine Learning Into Human Insights
0 likes1,083 views
This document discusses the strengths and weaknesses of machines versus humans in making insights from machine learning. It notes that while machines can perform cold calculations quickly and adjust models, they cannot understand outside context problems or human emotions. Humans are better at solving outside context problems, testing hunches, communicating insights, and providing customer service. The document argues that combining the strengths of machines and humans through data-informed decisions, not just data-driven ones, provides the best results.
IAB Canada Roadshow 2015: Turning Machine Learning Into Human Insights
- 1. Turning Machine Learning into Human Insights
- 3. VS.
- 4. Except …
- 12. Massive CTR! Geo-‐target Demographics CreaCve Huge Sales! Low CPA! Algorithm
- 13. 8 Reasons computers need humans 3 Sure-fire tactics for mind- blowing, KPIs! MagazineOctober, 2015 Is local better?
- 18. Outside context problem: When the context that created results is invisible to the machine -‐ Paul Goad
- 19. Results Time CompeEtor big sale Outside Context Problem
- 26. Overall campaign CTR: 0.18% Geo-‐fence tacCc CTR: 0.39%
- 29. N O W
- 45. Machines • Make cold, hard calculaEons with no emoEon • Sound, staEsEcal models with no fuzzy hunches • Make quick adjustments • MulEple variable analysis • Never goes to sleep and doesn’t take Eme off Humans • Are emoEonal and understand emoEon • BeXer at solving the outside context problem • Test hunches • Sweeping changes • Communicate • Are creaEve • Bring history • Provide customer service