main

# Sana Naik
# neo.csv = https://www.kaggle.com/datasets/sameepvani/nasa-nearest-earth-objects
dataset <- read.csv("neo.csv")

# View the data and plots
summary(dataset)

# Estimated size
library(ggplot2)
ggplot(dataset, aes(x = est_diameter_min, y = est_diameter_max)) + geom_point() + 
      labs(x = "Estimated Max Diameter", y = "Estimated Min Diameter",
      title = "Scatter Plot of Estimated Max vs. Min Diameter")

# Relative Velocity
hist(dataset$relative_velocity, main = "Histogram of Asteroid Relative Velocity", xlab = "Velocity")

# Miss Distance
hist(dataset$miss_distance, main = "Histogram of Asteroid Miss Distance", xlab = "Distance")

# Absolute Magnitude
hist(dataset$absolute_magnitude, main = "Histogram of Asteroid Absolute Magnitude", xlab = "Magnitude")

# Plot barplot of hazardous vs. non-hazardous asteroids
barplot(table(dataset$hazardous), main = "Hazardous vs. Non-hazardous Asteroids")

# Execute rpart, plot the decision tree
library(rpart)
tree <- rpart(hazardous ~ est_diameter_min + est_diameter_max + relative_velocity + miss_distance + absolute_magnitude,
              data = dataset, method="class")
library(rpart.plot)
rpart.plot(tree)

# Predict whether each asteroid will be classified as hazardous or not
dataset$predHazardous <- predict(tree, data=dataset, type = "class")

# Confusion Matrix
rpconfusionmatrix <- table(dataset$hazardous, dataset$predHazardous)
rpconfusionmatrix

# Naive Bayes
library(e1071)
bayesModel <- naiveBayes(hazardous ~ est_diameter_min + est_diameter_max + relative_velocity + miss_distance + absolute_magnitude,
                         data = dataset, method="class")

# Predict using the model
dataset$nbpredHazardous <- predict(bayesModel, dataset)

# Confusion matrix
nbconfusionmatrix <- table(dataset$hazardous, dataset$nbpredHazardous)
nbconfusionmatrix

# Compare accuracy
rpAccuracy <- sum(diag(rpconfusionmatrix))/sum(rpconfusionmatrix)
nbAccuracy <- sum(diag(nbconfusionmatrix))/sum(nbconfusionmatrix)
rpAccuracy
nbAccuracy