Want to Impress Me? Predict the Future, Not the Past
Niels Bohr was quoted in 1970 as saying, “Prediction is very difficult, especially about the future.”
So it is, but, of course, only predictions about the future have any value.
I appreciate the effort that goes into developing an algorithm to predict the success of something like a Broadway show, but this attempt shouldn’t be the basis for changing your marketing plans.
If you’re looking back at results, you can always make a model that appears to have predicted reality. Start with your actual data, and then keep tweaking the math ’til it works. This is especially true if you just look at 10 shows. (To be fair, this seems to be a project done more for fun than anything — so I’m not really criticizing, just using it to make a point.)
For a model to have predictive value, it must be able to say what’s going to happen to a show before it runs, and it must be able to do this reliably or, at least, significantly more often than not.
Till then, don’t redo your marketing plans based on this formula. And if you DO find such a formula, and it does work … you might want to keep it to yourself!