for the metrics chapter

[EricSchles]

https://en.wikipedia.org/wiki/Norm_(mathematics)

[EricSchles]

for metrics:

define precision as - of the things predicted as class A, the number of things that were of class A.

Trivial example - if you predict one thing as of class A and get it right, your precision is 100%.

define recall as true positive rate - the number of positive cases found of the total number of positive cases.

[EricSchles]

We can go one step further:

Precision - how sure are we about what we predicted?
Recall - how much should we trust that we captured a good number of the cases?

Or

Precision - how often we are right?
Recall - how often we detect the class?