library(e1071)
library(MLmetrics)
library(ROCR)

data = read.csv('../bayes.csv')
ytest = data[,1]
pred_m1 = data[,2]
pred_m2 = data[,3]

# Threshold 0.5
pred_m1[pred_m1 >= 0.5] = 1
pred_m1[pred_m1 < 0.5] = 0
m1 = prediction(pred_m1, ytest)
ROCcurve <- performance(m1, "tpr", "fpr")
plot(ROCcurve, col = "blue")
abline(0,1, col = "grey")
performance(m1, "auc")

pred_m2 = data[,3]
pred_m2[pred_m2 < 0.5] = 0
pred_m2[pred_m2 >= 0.5] = 1
ConfusionMatrix(pred_m2, ytest)
m2 = prediction(pred_m2, ytest)
# Define positive class when calling Rec/Prec/F1
Recall(pred_m2, ytest, '1')
Precision(pred_m2, ytest, '1')
F1_Score(pred_m2, ytest, '1')
ROCcurve2 <- performance(m2, "tpr", "fpr")
plot(ROCcurve2, col = "red")

# Get the AUC
performance(m2, "auc")