roc evaluation.Rmd

---
title: "R Notebook"
output:
  word_document:
    toc: yes
  html_document:
    df_print: paged
  html_notebook:
    number_sections: yes
    toc: yes
---
# Performance visualization with ROC
## Set-up
Clear the workspace and load packages. 
```{r}
rm(list = ls())
library(readxl)
library(tidyverse)
```

## Load final prediction csv of Xgboost, SVM, randomforest, classification tree, NNET model.
Let's load the dataset as xgboost.
```{r}
xgboost <- read.csv("roc_data/Xgboost.csv")
view(xgboost)
svm <- read.csv("roc_data/SVMPrediction.csv")
view(svm)
randomforest <- read.csv("roc_data/rf.csv")
view(randomforest)
tree1 <- read.csv("roc_data/tree1.csv")
view(tree1)
tree2 <- read.csv("roc_data/tree2.csv")
view(tree2)
temp <- read.csv("roc_data/temp1.csv")
view(temp)
```


## Plot ROC curves developed one by one.
```{r}
#install.packages("pROC")
library(pROC)

temp_roc<-roc(temp$real,temp$predict,auc=TRUE)
plot(temp_roc,print.auc=TRUE, print.auc.y=.31,col="black")

xgboost_roc<-roc(xgboost$y_real,xgboost$y_pred,auc=TRUE)
plot(xgboost_roc,print.auc=TRUE,print.auc.y=.23,col="blue",add=TRUE)

tree2_roc<-roc(tree2$y_real,tree2$y_pred,auc=TRUE)
plot(tree2_roc,print.auc=TRUE, print.auc.y=.39,col="purple",add=TRUE)

randomforest_roc<-roc(randomforest$y_real,randomforest$y_pred,auc=TRUE)
plot(randomforest_roc,print.auc=TRUE, print.auc.y=.47,col="green",add=TRUE)

svm_roc<-roc(svm$y_real,svm$y_pred,auc=TRUE)
plot(svm_roc,print.auc=TRUE, print.auc.y=.07,col="red",add=TRUE)

tree1_roc<-roc(tree1$y_real,tree1$y_pred,auc=TRUE)
plot(tree1_roc,print.auc=TRUE, print.auc.y=.15,col="orange",add=TRUE)

#add legend
par(mar=c(5.1,4.1,4.1,2.1), xpd=TRUE)
legend("bottomright", c("RandomForest","Classification Tree 2","Nnet","XGboost", "Classification Tree 1","SVM"), lty=1, 
    col = c("green","purple","black","blue","orange","red"), bty="n", inset=c(0,0))

```