Can someone elaborate more on the problem with "aggregating on percent success" part?
Why the underlying distribution could be a problem and why it makes sense to fix it just by extracting DOW and calculating success rates?
It's like saying profit rose from 100,000 to 200,000 in year 1 is a 100% increase and so does 1,000,000 to 2,000,000 in year 5. But the weight is different. Aggregating on the percent will treat these as equal outcomes but it's not. At least this is what I gathered. Maybe the course creator can cmiiw.