DataFramed #62 From Predictions to Decisions
May 17, 2021
Dan Becker, CEO of decision.ai and founder of Kaggle Learn, shares insights on bridging decision sciences with machine learning. He discusses his journey from Kaggle competitions to real-world applications, emphasizing the importance of empathy in aligning AI with business goals. Dan highlights challenges in integrating decision AI into climate change projects and the necessity of ongoing monitoring for model accuracy. He advocates for data scientists to communicate effectively with leadership to ensure their work resonates with business priorities.
AI Snips
Chapters
Transcript
Episode notes
Early ML Failure
- Dan Becker's first data science venture involved using neural networks to optimize eBay listings for retailers.
- The project failed, leading him to believe machine learning wouldn't succeed.
Kaggle Comeback
- Dan re-entered data science through a Kaggle competition, initially performing poorly.
- Dedicated practice led him to second place, sparking a consulting career.
Iterative Improvement
- Iteratively improve your models, seeking feedback and learning from mistakes.
- Engage with data science communities and learn through conversations.