BH <- read.csv("BostonHousing Long.csv")
BH_subset <- BH[, c("CHAS", "RAD", "CAT.MEDV")]
BH_subset[] <- lapply(BH_subset, factor)
set.seed(1234)
train_index <- sample(1:nrow(BH_subset), 0.6 * nrow(BH_subset))
train_data <- BH_subset[train_index, ]
test_data <- BH_subset[-train_index, ]
library(e1071)
Naive_Bayes <- naiveBayes(CAT.MEDV ~ ., data = train_data)
table(train_data$CHAS, train_data$CAT.MEDV)
##
## 0 1
## 0 121 161
## 1 7 14
highCHAS <- sum(train_data$CHAS == 1 & train_data$CAT.MEDV == 1)
num_CHAS <- sum(train_data$CHAS == 1)
highCHAS <- highCHAS / num_CHAS
highCHAS
## [1] 0.6666667
value_not_CHAS <- sum(train_data$CHAS == 0 & train_data$CAT.MEDV == 1)
num_not_CHAS <- sum(train_data$CHAS == 0)
value_given_not_CHAS <- value_not_CHAS / num_not_CHAS
value_given_not_CHAS
## [1] 0.570922
predictions <- predict(Naive_Bayes, newdata = test_data)
predictions[10]
## [1] 1
## Levels: 0 1
test_data[10,]
## CHAS RAD CAT.MEDV
## 16 0 4 0
library(caret)
## Loading required package: ggplot2
## Loading required package: lattice
conf_matrix <- confusionMatrix(data = as.factor(predictions), reference = as.factor(test_data$CAT.MEDV))
conf_matrix
## Confusion Matrix and Statistics
##
## Reference
## Prediction 0 1
## 0 32 16
## 1 55 100
##
## Accuracy : 0.6502
## 95% CI : (0.5804, 0.7157)
## No Information Rate : 0.5714
## P-Value [Acc > NIR] : 0.01337
##
## Kappa : 0.2435
##
## Mcnemar's Test P-Value : 6.49e-06
##
## Sensitivity : 0.3678
## Specificity : 0.8621
## Pos Pred Value : 0.6667
## Neg Pred Value : 0.6452
## Prevalence : 0.4286
## Detection Rate : 0.1576
## Detection Prevalence : 0.2365
## Balanced Accuracy : 0.6149
##
## 'Positive' Class : 0
##
error_rate <- round(1 - conf_matrix$overall['Accuracy'], 2)
error_rate
## Accuracy
## 0.35
sensitivity <- round(conf_matrix$byClass['Sensitivity'], 2)
sensitivity
## Sensitivity
## 0.37
specificity <- round(conf_matrix$byClass['Specificity'], 2)
specificity
## Specificity
## 0.86
mode_value <- as.numeric(names(sort(table(train_data$CAT.MEDV), decreasing = TRUE)[1]))
mode_value
## [1] 1
library(pROC)
## Type 'citation("pROC")' for a citation.
##
## Attaching package: 'pROC'
## The following objects are masked from 'package:stats':
##
## cov, smooth, var
roc_curve <- roc(response = test_data$CAT.MEDV, predictor = as.numeric(predictions))
## Setting levels: control = 0, case = 1
## Setting direction: controls < cases
auc <- auc(roc_curve)
auc
## Area under the curve: 0.6149
roc_curve <- roc(response = test_data$CAT.MEDV, predictor = as.numeric(predictions))
## Setting levels: control = 0, case = 1
## Setting direction: controls < cases
plot(roc_curve, main = "ROC Curve", col = "blue")
roc_curve <- roc(response = test_data$CAT.MEDV, predictor = as.numeric(predictions))
## Setting levels: control = 0, case = 1
## Setting direction: controls < cases
plot(roc_curve, main = "Lift Chart", col = "blue", print.auc = TRUE, print.auc.x = 0.5, print.auc.y = 0.5, lwd = 2)
