What can go wrong with machine learning? While at NDC in Toronto, Richard chatted with Megan Robertson about her experience with machine learning projects, often using retail datasets, and where they can go wrong. Megan talks about getting clear expectations and metrics for projects, so you know when you succeed, but then digs into the specifics of problems in machine learning, such as overfitting on test data. Your results are only as good as the data you put in, so a lot of focus goes into building good sets, carefully developing the model with those sets, and using techniques like cross-validation to ensure the model is behaving appropriately. There's a lot that can go wrong, but the results with an effective model can be very powerful - it is worth the effort!