Roadmap: How to Learn Machine Learning in 6 Months
- 1. ROADMAP:SIX MONTHS TO MACHINE LEARNING ZACHARIAH MILLER - 5/20/17
- 2. WHO AM I?AND WHY SHOULD YOU LISTEN TO ME? This is the STAR Detector at Brookhaven National Laboratory.
- 3. IF YOU ONLY REMEMBER ONE THING FROM THIS TALK
- 4. JUST BUILD SOMETHING WITH DATA AND EXPECT TO SUCK AT IT, FOR A WHILE
- 6. My First Data Project Edited. But the first one made the plots in my thesis.
- 8. MATH
- 9. MATH & MACHINE LEARNING ▸ Linear Algebra ▸ Calculus ▸ Statistics ▸ Probability
- 10. MATH & MACHINE LEARNING ▸ MIT Linear Algebra Open Course ▸ MIT Calculus Open Course ▸ MIT Stats and Probability Course
- 11. VS Python R
- 12. VS BOTH ARE GOOD PICK ONE AND LEARN
- 14. THE MACHINE LEARNING PIPELINE ▸ Find Data (Webscrape, APIs, CSVs) ▸ Clean the data (remove NaNs and Infinities, should that be a string? Probably not, maybe I can categorize it…) YOU’LL SPEND A LOT OF TIME IN THESE SECTIONS. THAT’S NORMAL.
- 15. THE MACHINE LEARNING PIPELINE ▸ Find Data (Webscrape, APIs, CSVs) ▸ Clean the data (remove NaNs and Infinities, should that be a string? Probably not, maybe I can categorize it…) ▸ Choose and tune your algorithm ▸ Visualize results GOOGLE AND STACK OVERFLOW ARE YOUR FRIENDS
- 16. GOOGLE AND STACK OVERFLOW ARE YOUR FRIENDS THE MACHINE LEARNING PIPELINE ▸ Find Data (Webscrape, APIs, CSVs) ▸ Clean the data (remove NaNs and Infinities, should that be a string? Probably not, maybe I can categorize it…) ▸ Choose and tune your algorithm ▸ Visualize results
- 18. SO… SIX MONTHS? ▸ Learn the math. (2 - 3 months) ▸ Learn the programming language (1 month) ▸ Machine learning tutorials and test projects (1 - 2 months) ▸ Short term passion projects (1+ month)
- 19. DATA SCIENCE AND MACHINE LEARNING CLASSES/BOOTCAMPS
- 20. SOME LAST NOTES ▸ You’re going to fail. A lot. It’s software… so who cares. ▸ Your models may not be predictive. That’s a result. Null is just as good as non- null if you did it right. ▸ Track your projects on GitHub and write up your results.
- 21. THANKS!LET’S CHAT. I’D LOVE TO TALK ABOUT PROJECTS YOU’RE CONSIDERING.
- 22. PHOTO CREDITS ▸ https://c1.staticflickr.com/4/3310/3313585315_4874a81f77_b.jpg (STAR Detector) ▸ https://c1.staticflickr.com/6/5531/9638435181_7e3e44c2b8_b.jpg (Highway) ▸ https://www.mathworks.com/matlabcentral/mlc-downloads/ downloads/submissions/35389/versions/1/screenshot.png (Gradient Descent) ▸ http://blog.yhat.com/static/img/roc-auc.png (AUC) ▸ http://scikit-learn.org/stable/tutorial/machine_learning_map/ (SkLearn Map)