Bradley Hof
For this shiny application, we are looking for an easy to use statistical model to predict future NFL scores and winners. We would like to investigate the following co-variates
The data for this application contain scores on all 2014 NFL games up to week 11
## 'data.frame': 162 obs. of 9 variables:
## $ Week : int 1 1 1 1 1 1 1 1 1 1 ...
## $ home.team: Factor w/ 32 levels "49ERS","BEARS",..: 2 5 7 8 10 12 13 14 15 18 ...
## $ away.team: Factor w/ 32 levels "49ERS","BEARS",..: 4 11 21 9 31 1 22 17 27 23 ...
## $ Line : num -6.5 -8 -4.5 -3 -3 3.5 3.5 -10 3 -6.5 ...
## $ Surface : Factor w/ 3 levels "Dome","Grass",..: 2 2 2 2 2 3 2 2 1 3 ...
## $ temp : num 77 70 79 72 77 84 89 80 72 80 ...
## $ condition: Factor w/ 22 levels "/53F CHANCE SHOWERS",..: 18 18 16 5 6 18 6 18 5 18 ...
## $ HS : int 20 31 14 18 10 17 33 34 37 19 ...
## $ RS : int 23 24 20 17 26 28 20 17 34 14 ...
In the plots below, you can see that a home team score is not affected by the surface of the field. However, the surface does have a slight impact on the road team's score.
We used a simple linear model for this game predictor.
Try the NFL Prediction App today!!
load("C:\\projects\\coursera\\nfl\\NFLdata.rda")
homeFit <- lm(RS ~ home.team + away.team + Line + Surface + temp + condition, data=nfl)
plot(homeFit)
