I can't speak to how good this particular example is, or any other similar solutions either. However, if it can predict better than 50/50 (e.g. 60-70%) then th

I can't speak to how good this particular example is, or any other similar solutions either. However, if it can predict better than 50/50 (e.g. 60-70%) then that can certainly be useful. Take digital advertising for example, many advertisers want to target either female or male demographic exclusively. 60% might not be enough to sell adverts as "this will only be shown to men", but it might help you behind the scenes serve the most relevant adverts to people, potentially increasing click-through rate, and therefore the value of your service. Sure there are other examples, but my mind is stuck in digital banner advertising this afternoon.