Machine Learning success and failure cases: from Metallurgy and HR to online dating and e-commerce
Key takeaways include:
- Learn why you should make sure the problem really exists before trying to solve it (it is often not the case, and often there is nothing to optimize), or you'll waste time
- Learn why you should be sure to formulate the technical problem in a meaningful way from the business problem, or you'll waste time and not get the expected result in the end.
- Learn why you should be sure to formulate the success criteria and the experiment to test the hypothesis in the beginning, or you'll waste a lot of time when there's already a solution at hand, but you can't test whether it's really a better solution.