Feature Analysis.Rmd

---
title: "Analysis of Model Features"
output: html_notebook
---

Exploratory analysis of all features used in the creation of the NFL Betting Model.

```{r}
nflModelData <- read.csv("/Users/anguslai/Documents/HBA2/Intro to Data Sci/Group Project/Data/final_v3.csv", head=TRUE)
head(nflModelData)
```

Charts 

```{r}
plot(nflModelData$win_loss,xlab="Team",ylab="Number of Wins",main="Team Wins by Classification")
print(paste0("Teams that were the favourite won ", (round(sum(nflModelData$win_loss == "fav")/nrow(nflModelData),3)*100),"% of games"))
```

```{r}
IsFav = ("fav" == nflModelData$win_loss)
IsHomeFav = ("home_fav" == nflModelData$home_fav)
print(paste0("Teams that were the favourite to win and played at home won ",round(sum(IsFavAndHomeFav = (IsFav & IsHomeFav))/sum(IsHomeFav)*100,1),"% of games."))
```


```{r}
library(ggplot2)
ggplot(nflModelData, aes(x=nflModelData$schedule_season, y=nflModelData$over_under_line)) + geom_point(color="#69b3a2") + geom_smooth(method=lm , color="red", se=FALSE) + labs(title="Over/Under Line from 1985 - 2018", x ="NFL Season", y = "Over/Under Line")
```

Correlation

```{r}
corrData <- nflModelData

corrData$schedule_playoff <- as.character(corrData$schedule_playoff)
corrData$schedule_playoff[corrData$schedule_playoff=='FALSE'] = 0
corrData$schedule_playoff[corrData$schedule_playoff=='TRUE'] = 1
corrData$schedule_playoff <- as.numeric(corrData$schedule_playoff)

corrData$win_loss <- as.character(corrData$win_loss)
corrData$win_loss[corrData$win_loss=='underdog'] = 0
corrData$win_loss[corrData$win_loss=='fav'] = 1
corrData$win_loss <- as.numeric(corrData$win_loss)

corrData$home_fav <- as.character(corrData$home_fav)
corrData$home_fav[corrData$home_fav=='home_underdog'] = 0
corrData$home_fav[corrData$home_fav=='home_fav'] = 1
corrData$home_fav <- as.numeric(corrData$home_fav)

corrData <- subset(corrData, select = -c(team_favorite_id))

head(corrData)
```

```{r}
round(cor(corrData[,1:9]),3)
```


```{r}
library(corrplot)
corrData <- as.matrix(corrData)
corrplot(cor(corrData))
```