Assignment

#Load required libraries library(imager) library(caret) library(magick) library(tidyverse) library(skimr) library(DataExplorer) library(ggplot2) library(neuralnet) library(dplyr) library(nnet)

Step 1: Load images, convert to grayscale, and create dataset

image_folder <- (“C:/Users/Kavyadinesh/Downloads/New”) # Function to convert an image to grayscale convert_to_grayscale <- function(image_path, save_path) { # Read the image img <- image_read(image_path) # Convert to grayscale grayscale_img <- image_convert(img, colorspace = ‘gray’) # Save the grayscale image image_write(grayscale_img, save_path) }

Loop through each subfolder and each image in the main directory

subdirs <- list.dirs(image_folder, full.names = TRUE, recursive = TRUE)

for (subdir in subdirs) { image_files <- list.files(subdir, pattern = “\.(jpg|png)$”, full.names = TRUE)

for (file in image_files) { # Define where to save the grayscale image save_path <- file.path(subdir, paste0(“grayscale_”, basename(file)))

Convert the image to grayscale and save it

convert_to_grayscale(file, save_path)

Print a message indicating the conversion

cat("Converted", basename(file), "to grayscale and saved as", save_path, "\n")

} }

Define the directory containing your grayscale images

image_folder <- (“C:/Users/Kavyadinesh/Downloads/Grayscale”)

Function to convert an image to a vector of pixel values

image_to_vector <- function(image_path) { # Read the image img <- image_read(image_path) # Convert image to grayscale (if not already grayscale) grayscale_img <- image_convert(img, colorspace = ‘gray’) # Resize the image to a fixed size (optional, e.g., 100x100 pixels) resized_img <- image_resize(grayscale_img, “100x100!”) # Convert the image to a matrix of pixel values img_matrix <- as.integer(image_data(resized_img)[1,,]) # Flatten the matrix into a vector as.vector(img_matrix) }

Initialize an empty list to store image data

image_data_list <- list() image_labels <- c()

Loop through each subfolder and each image in the main directory

subdirs <- list.dirs(image_folder, full.names = TRUE, recursive = TRUE) for (subdir in subdirs) { image_files <- list.files(subdir, pattern = “\.(jpg|png)$”, full.names = TRUE) # Get the label (subfolder name) label <- basename(subdir) for (file in image_files) { # Convert the image to a vector of pixel values image_vector <- image_to_vector(file) # Add the image vector to the list image_data_list[[length(image_data_list) + 1]] <- image_vector # Store the label for this image image_labels <- c(image_labels, label) } }

Combine the image vectors into a data frame

image_data_df <- do.call(rbind, image_data_list) image_data_df <- as.data.frame(image_data_df) # Add the labels as a new column image_data_df$label <- image_labels

Define the path where you want to save the CSV file

csv_file_path <- (“C:/Users/Kavyadinesh/Downloads/Grayscale/data.csv”)

Save the data frame as a CSV file

write.csv(image_data_df, csv_file_path, row.names = FALSE)

Confirmation message

cat(“Dataset has been saved as a CSV at”,csv_file_path)

#PREPROCESSIN AND EDA # Load the CSV file data <- read.csv(“C:/Users/Kavyadinesh/Downloads/Grayscale/data.csv”)

View the first few rows of the dataset

head(data)

Get the structure of the dataset

str(data)

Get a summary of the dataset

summary(data)

Check for missing values

sum(is.na(data)) # Total missing values

Count of missing values by column

colSums(is.na(data))

Remove rows with missing values

data_clean <- na.omit(data) data_clean

Check for duplicates

duplicated_rows <- data[duplicated(data), ]

Remove duplicates

data_clean <- data %>% distinct()

Convert character variables to factors

data_clean <- data_clean %>% mutate(across(where(is.character), as.factor))

View the distribution of categorical variables

table(data_clean$your_categorical_column)

#PERFORMING LOGISTIC REGRESSION data\(label <- as.factor(data\)label) data$label

Split the data into training and test sets (70% training, 30% testing)

set.seed(123) # Set seed for reproducibility sample <- sample(c(TRUE, FALSE), nrow(data), replace = TRUE, prob = c(0.7, 0.3)) train_data <- data[sample, ] test_data <- data[!sample, ]

str(train_data)

Fit logistic regression model

logistic_model <- glm(label ~ ., data = train_data, family = binomial)

Summary of the logistic regression model

summary(logistic_model)

#K-MEANS CLUSTERING

library(tidyverse) names(data) clustering_data <- data %>% select(V1, V2, V3)

Set seed for reproducibility

set.seed(123)

Perform K-means clustering

kmeans_result <- kmeans(clustering_data, centers = 3)

View the clustering result

print(kmeans_result)

Add cluster assignments to the original dataset

data\(cluster <- as.factor(kmeans_result\)cluster)

View first few rows with cluster labels

head(data)

#NEURAL NETWORK # NEURAL NETWORK WITH 1 HIDDEN LAYER 5 NEURONS # Set seed for reproducibility set.seed(123)

Create a sample dataset

data <- data.frame( feature1 = rnorm(100), feature2 = rnorm(100), label = as.factor(sample(c(0, 1), 100, replace = TRUE)) # Binary target )

Split data into training and testing sets

train_index <- createDataPartition(data$label, p = 0.8, list = FALSE) train_data <- data[train_index, ] test_data <- data[-train_index, ]

Scale the predictor variables (excluding the target variable)

train_data_scaled <- as.data.frame(scale(train_data[, -3])) # Scale features only train_data_scaled\(label <- train_data\)label # Add label back

Fit the neural network model with 1 hidden layer and 5 neurons

nn_model <- neuralnet(label ~ ., data = train_data_scaled, hidden = 5, linear.output = FALSE)

NEURAL NETWORK WITH 1 HIDDEN LAYER 10 NEURONS

set.seed(123)

Generate data

data <- data.frame( feature1 = rnorm(100), feature2 = rnorm(100), label = as.factor(sample(c(0, 1), 100, replace = TRUE)) # Binary target )

Split data into training and testing sets

train_index <- createDataPartition(data$label, p = 0.8, list = FALSE) train_data <- data[train_index, ] test_data <- data[-train_index, ]

Scale the predictor variables (excluding the target variable)

train_data_scaled <- as.data.frame(scale(train_data[, -3])) # Scale features only train_data_scaled\(label <- train_data\)label # Add label back

Fit the neural network with 10 neurons in 1 hidden layer

nn_model <- neuralnet(label ~ ., data = train_data_scaled, hidden = 10, linear.output = FALSE)

NEURAL NETWORK WITH 1 HIDDEN LAYER 20 NEURONS

set.seed(123)

Generate data

data <- data.frame( feature1 = rnorm(100), feature2 = rnorm(100), label = as.factor(sample(c(0, 1), 100, replace = TRUE)) # Binary target )

Split data into training and testing sets

train_index <- createDataPartition(data$label, p = 0.8, list = FALSE) train_data <- data[train_index, ] test_data <- data[-train_index, ]

Scale the predictor variables (excluding the target variable)

train_data_scaled <- as.data.frame(scale(train_data[, -3])) # Scale features only train_data_scaled\(label <- train_data\)label # Add label back

Fit the neural network with 1 hidden layer and 20 neurons

nn_model <- neuralnet(label ~ ., data = train_data_scaled, hidden = 20, linear.output = FALSE)