ctosummitdatasciencekeynote-150516193652-lva1-app6891.pdf

This 2015 keynote at the CTO Summit on Data Science emphasizes that the key to effective data science is identifying and framing the right problem. It discusses the importance of explainability in models and the iterative nature of building machine learning models. It highlights the need for optimizing the speed of learning and conducting experiments wisely in order to achieve better results.

Transcript

1. Data Science: A Mindset for Productivity Daniel Tunkelang @dtunkelang Daniel
2. tl;dr The most important part of data science is picking the right problem and figuring out how to frame it.
3. We’re all technologists, right?
4. But nobody knows everything.* Class HashMap<K,V> java.lang.Object java.util.AbstractMap<K,V> java.util.HashMap<K,V> Type Parameters: K - the type of keys maintained by this map V - the type of mapped values All Implemented Interfaces: Serializable, Cloneable, Map<K,V> *Except Jeff Dean.
5. Math and computer science matter…
6. But you have to solve the right problem.
7. Stay friends with your exes. explain express experiment
8. Data science is a mindset. Explain Iterate using explainable models. Express Model your utility and inputs. Experiment Optimize for speed of learning.
9. Explain
10. With apologies to the little prince.
11. Deep learning is the new black.
12. But accuracy isn’t everything.
13. The importance of being explainable. • Algorithms can protect you from overfitting, but they can’t protect you from the biases you introduce. • Introspection into your models and features makes it easier for you and others to debug them. • Especially if you don’t completely trust your objective function or representativeness of your training data.
14. Linear models? Decision trees? • Linear regression and decision trees favor explainability over accuracy, compared to more sophisticated models. • But size matters. If you have too many features or too deep a decision tree, you lose explainability. • You can always upgrade to a more sophisticated model when you trust your objective function and training data. • Build a machine learning model is an iterative process. Optimize for the speed of your own learning.
15. Express
16. Machine learning for dummies. • Define objective function. • Collect training data. • Build models. • Profit!
17. You only improve what you measure. Clicks? Actions? Outcomes?
18. Sometimes accuracy is complicated.
19. What’s your error function?
20. Consider stratified sampling.
21. Experiment
22. How to find your prince. You have to kiss a lot of frogs to find one prince. So how can you find your prince faster? By finding more frogs and kissing them faster and faster. -- Mike Moran
23. Think like an economist. Yesterday Experiments are expensive, choose hypotheses wisely. Today Experiments are cheap, do as many as you can!
24. But don’t forget you’re a scientist.
25. Optimize for the speed of learning.
26. Test one variable at a time. • Autocomplete • Entity Tagging • Vertical Intent • # of Suggestions • Suggestion Order • Language • Query Construction • Ranking Model
27. tl;dr The most important part of data science is picking the right problem and figuring out how to frame it.
28. Daniel Tunkelang dtunkelang@gmail.com https://linkedin.com/in/dtunkelang @dtunkelang