Naive Bayes Classifier in R

pima.R

require(e1071)
pima_data <- read.csv(file = "pima.data", header=TRUE)

#load data into data frame

names(pima_data) <- c(1:9) 

#   1. Number of times pregnant
#   2. Plasma glucose concentration a 2 hours in an oral glucose tolerance test
#   3. Diastolic blood pressure (mm Hg)
#   4. Triceps skin fold thickness (mm)
#   5. 2-Hour serum insulin (mu U/ml)
#   6. Body mass index (weight in kg/(height in m)^2)
#   7. Diabetes pedigree function
#   8. Age (years)
#   9. Class variable (0 or 1)

training_portion <- 0.8

h <- c(train = training_portion, test = 1-training_portion)
g <- sample(cut(seq(nrow(pima_data)), nrow(pima_data)*cumsum(c(0,h)), labels = names(h)))
epoch_data <- split(pima_data, g)
train_data <- epoch_data$train
test_data <- epoch_data$test
train_class_0 <- train_data[train_data$'9'==0,]
train_class_1 <- train_data[train_data$'9'==1,]

train_class_0_var_by_par <- lapply(train_class_0[,c(1:8)], var)
train_class_1_var_by_par <- lapply(train_class_1[,c(1:8)], var)

train_class_0_mean_by_par <- lapply(train_class_0[,c(1:8)], mean)
train_class_1_mean_by_par <- lapply(train_class_1[,c(1:8)], mean)

p_label_0 = nrow(train_class_0) / nrow(train_data)
p_label_1 = 1.0 - p_label_0

p_condition_class <- function (attr, val, class) {
	if (attr < 1 || attr > 8) {
		return(-1)
	}
	else {
		var_res <- 0.0
		mean_res <- 0.0
		if (class == 1) {
			var_res <- train_class_1_var_by_par[[toString(attr)]]
			mean_res <- train_class_1_mean_by_par[[toString(attr)]]
		}
		else {
			if (class == 0) {
				var_res <- train_class_0_var_by_par[[toString(attr)]]
				mean_res <- train_class_0_mean_by_par[[toString(attr)]]	
			}
			else {
				return(-1)
			}	
		}
		
		ret <- (1.0 / (sqrt(2.0 * pi * var_res))) * exp(-1 * (val - mean_res)^2 / (2 * var_res))
		return(ret)
	}
}

naive_bayes_formula <- function (row) {
	condition_prob_0 <- 1.0
	for (i in c(1:8)) {
		condition_prob_0 <- condition_prob_0 * (p_condition_class(i,row[toString(i)],0))
	}
	condition_prob_0 <- condition_prob_0 * p_label_0
	
	condition_prob_1 <- 1.0
	for (i in c(1:8)) {
		condition_prob_1 <- condition_prob_1 * (p_condition_class(i,row[toString(i)],1))
	}
	condition_prob_1 <- condition_prob_1 * p_label_1
	
	if (condition_prob_0 < condition_prob_1) {
		return(1)
	}
	else {
		return(0)
	}
}

accuracy = 0.0
for (i in c(1:nrow(test_data))) {
	estimated_label <- (naive_bayes_formula(test_data[i,]))
	if (estimated_label == test_data[i, '9']) {
		accuracy = accuracy + 1.0
	}
}
print (accuracy/nrow(test_data))