How to Prioritize Machine Learning Initiatives

The best problems to solve machine learning is our one that has high business impact as well as technical feasibility. And also use a scorecard when prioritizing multiple initiatives this is a really great way to get all your thoughts down on paper and assess what initiatives make the most sense. It's important to consider and plan for edge cases that I mentioned so a lot of the time this is where you can get in trouble if you're not considering the cost of error with these edge cases. Try to bring up where the machine learning product can add value earlier in the life cycle by doing offline analysis, as well as doing smaller pilots when you are watching.