Machine Learning The Hard Way

https://www.youtube.com/watch?v=xeAB10QgDW8

• time(working with data) > time(optimizing algorithms)
• understanding your data. garbage in, garbage out.
• formatting your data. munging. (numpy, scikit-learn)
• writing your own solution is a good indicator that you don't know what you're doing.
• pre-processing your data. normalize.
• know your goal. gap between real probability and odds implied probability.
• choosing an algorithm.
• read the algorithm docs.
• testing results because you're probably wrong. (cross_validation, shuffle data)
• interpreting results. don't bet the house.
• share your knowledge. you know more than you think.
• keep learning. you know less than you think.