intro: More "Your turn" in slides introducing metrics

[simonpcouch]

trafficstars

I think it might be interesting for folks to try and "derive" accuracy based on TP / FP/ FN / TN, i.e.

Here are four quantities you have access to. What is the simplest way to combine them to determine how often the model is right?

Then, the slide on pitfalls on accuracy, then, another "derive" a metric to balance classes.

This might help motivate the three default metrics a bit more.