# Machine Learning for NFL Analysis, 2017 Week 2

I finally finished the weekly prediction functions, and the resulting graphic is much more legible. A couple notes:

- The scale is from 0–1: a confidence of 0 indicates that the confidence that the model has in its prediction of the winner of the game is nonexistent. Another way to think about this is the model predicts that either team has a 50% chance of winning. A score of 1 indicates that the model predicts that its pick has a 100% chance of winning. How this determination is made is dependent on the type of model itself.
- Keep in mind that these models are overall just slightly above 50% accurate! What this means is that a model’s confidence in its prediction can only be a percentage of its overall accuracy. So even if a 55% accurate model predicts a winner with 100% confidence, there is still only about a 55% chance of that prediction being accurate.
- Oh also apparently the Chargers are abbreviated as SD, and the Rams as LA. Sorry for the confusion.

That said, on to the week 2 prediction results:

Judging by aggregate prediction, the winners with the top confidence scores are:

**CLE**over BAL**NE**over NO**PIT**over MIN**DEN**over DAL**LAR**over WAS