Boston Consulting Group Interview Question

They challenged my classification approach and asked whether the problem should instead be treated as a forecasting problem. The follow-up focused on how the ML workflow would change, especially train/test splitting, time dependency considerations, and evaluation for time-series data.