#logistic Regression
# Install the 'caret' package (only run once; comment out if already installed)
install.packages("caret")

# Load the 'caret' package for machine learning utilities
library(caret)

# Load the built-in iris dataset
data(iris)

# Convert the problem into binary classification:
# Setosa (1) vs Non-Setosa (0)
iris$Label <- ifelse(iris$Species == "setosa", 1, 0)

# Remove the original Species column as it's no longer needed
iris <- iris[, -5]

# Set seed for reproducibility
set.seed(123)

# Split the data: 80% for training and 20% for testing
idx <- createDataPartition(iris$Label, p = 0.8, list = FALSE)
train <- iris[idx, ]   # Training set
test <- iris[-idx, ]   # Test set

# Train a logistic regression model using the training data
model <- glm(Label ~ ., data = train, family = "binomial")

# Predict probabilities on the test set and convert to class labels (1 or 0)
pred <- ifelse(predict(model, test, type = "response") > 0.5, 1, 0)

# Generate a confusion matrix to evaluate model performance
conf <- confusionMatrix(factor(pred), factor(test$Label))

# Display evaluation metrics
cat("Precision:", round(conf$byClass["Precision"], 2), "\n")
cat("Recall:", round(conf$byClass["Recall"], 2), "\n")
cat("F1-score:", round(conf$byClass["F1"], 2), "\n")