1. Load the Titanic dataset using read.

csv(), explore it with str(), summary(), and

head(). Identify numerical/categorical variables and handle missing values by
imputing with mean, median, or mode. Convert categorical data into factors, scale
numerical features, engineer new features, and save the preprocessed data using
write.csv().

# Load data
titanic <- read.csv("titanic.csv")

# Explore data
str(titanic)
summary(titanic)
head(titanic)

# Handle missing values

titanic$Age[is.na(titanic$Age)] <- mean(titanic$Age, na.rm = TRUE)
titanic$Embarked[is.na(titanic$Embarked)] <- "S"

# Convert to factors
titanic$Sex <- as.factor(titanic$Sex)
titanic$Embarked <- as.factor(titanic$Embarked)

# Scale numerical features

titanic$Fare <- scale(titanic$Fare)
titanic$Age <- scale(titanic$Age)

# Engineer new feature: FamilySize

titanic$FamilySize <- titanic$SibSp + titanic$Parch + 1

# Save preprocessed data

write.csv(titanic, "titanic_cleaned.csv", row.names = FALSE)

2. Use Esquisse to visualize the mtcars dataset: 1.Create a histogram/density plot for
mpg (Miles per Gallon). 2. Compare the number of automatic vs. manual cars
using a bar plot of am.3.Show the relationship between cyl (Cylinders) and hp
(Horsepower) using a box/scatter plot. 4.Visualize mpg vs. wt (Weight) with color
for cyl. 5.Create a bar plot showing car counts by gear (Transmission gears).
Export and save the R code.

install.packages("esquisse")
library(esquisse)

# Load mtcars
data("mtcars")

# Launch Esquisse GUI

 esquisse::esquisser()

3. Load a mtcars dataset, identify numerical and categorical variables, and visualize
distributions using box plots, histograms, and violin plots. Use scatter plots with
trend lines to analyze relationships. Create facets, bar plots, and heatmaps to
explore patterns and correlations. Provide insights on distributions, outliers, and
variable relationships.

library(ggplot2)
library(vioplot)
data("mtcars")

# Boxplot
boxplot(mtcars$mpg, main="MPG Boxplot")

# Histogram
hist(mtcars$hp, main="HP Histogram", col="lightblue")

# Violin plot
vioplot::vioplot(mtcars$wt, names = "Weight")

# Scatter plot with regression line

ggplot(mtcars, aes(x=mpg, y=wt)) +
geom_point() + geom_smooth(method="lm")

# Correlation heatmap
heatmap(cor(mtcars), main="Correlation Heatmap")

4. Develop a Shiny app using the iris dataset with three features: (1) Select a
numerical and categorical variable to display summary statistics. (2) Choose two
numerical variables for a scatter plot, colored by a categorical variable. (3)
Generate a box plot for a selected numerical and categorical variable.

library(shiny)
data(iris)

ui <- fluidPage(
titlePanel("Iris Shiny App"),
selectInput("num", "Numerical Variable:", choices=names(iris)[1:4]),
selectInput("cat", "Categorical Variable:", choices=c("Species")),
verbatimTextOutput("summary"),
selectInput("x", "X-axis Variable:", choices=names(iris)[1:4]),
selectInput("y", "Y-axis Variable:", choices=names(iris)[1:4]),
plotOutput("scatter"),
plotOutput("boxplot")
)
 server <- function(input, output) {
output$summary <- renderPrint({
summary(iris[[input$num]])
})

output$scatter <- renderPlot({

plot(iris[[input$x]], iris[[input$y]], col=iris$Species, pch=19)
})

output$boxplot <- renderPlot({

boxplot(iris[[input$num]] ~ iris[[input$cat]], main="Boxplot")
})
}

shinyApp(ui = ui, server = server)

5. Analyze outliers in the mtcars dataset using IQR and Z-score methods. Compute
probabilities of selecting outlier cars based on mpg, hp, and wt. Explore
conditional probabilities, independence, and expected outliers. Examine
distribution shapes, skewness, correlations, and compare different outlier
detection methods to assess consistency and insights.

library(e1071)
data("mtcars")

# IQR method
Q1 <- quantile(mtcars$mpg, 0.25)
Q3 <- quantile(mtcars$mpg, 0.75)
IQR_val <- Q3 - Q1
outliers_iqr <- which(mtcars$mpg < (Q1 - 1.5 * IQR_val) | mtcars$mpg > (Q3 + 1.5 *
IQR_val))

# Z-score method
z_scores <- scale(mtcars$mpg)
outliers_z <- which(abs(z_scores) > 3)

# Print outliers
mtcars[outliers_iqr, ]
mtcars[outliers_z, ]

# Skewness
skewness(mtcars$mpg)
6. Perform correlation analysis on the mtcars dataset to evaluate relationships
between mpg and other features using Pearson and Spearman methods. Identify
key influencing variables and visualize insights using heatmaps, scatter plots, box
plots, bar plots, and clustering dendrograms, ensuring a comprehensive
understanding of mpg dependencies and trends.
data("mtcars")

data("mtcars")

# Pearson and Spearman

cor(mtcars$mpg, mtcars$hp, method = "pearson")
cor(mtcars$mpg, mtcars$hp, method = "spearman")

# Correlation heatmap
heatmap(cor(mtcars), main = "Correlation Heatmap")

# Dendrogram clustering
dist_m <- dist(mtcars)
hcl <- hclust(dist_m)
plot(hcl)

7. Analyze sales trends over time using scatter plots and regression models. Fit linear
and quadratic regression, compare their R-squared and RMSE values, and
interpret results. Identify the best-fitting model and predict sales for the next 6
months (Months: 61-66) to assess future trends and decision-making insights.

set.seed(123)
months <- 1:60
sales <- 100 + 2*months + rnorm(60, 0, 10)
data <- data.frame(months, sales)

# Linear regression
model1 <- lm(sales ~ months, data)
summary(model1)

# Quadratic regression
model2 <- lm(sales ~ months + I(months^2), data)
summary(model2)

# Predict next 6 months

future <- data.frame(months = 61:66)
predict(model2, newdata = future)
8. A medical researcher is studying the relationship between various blood test
parameters, such as glucose levels, cholesterol, and blood pressure, to identify
potential risk factors for diabetes. Compute the Pearson and Spearman
correlation coefficients for a given dataset and analyze the relationships between
variables. Explain how correlation matrices assist in feature selection and why
highly correlated features might lead to redundancy in predictive models. Using
linear algebra, describe how the correlation matrix can be decomposed using
eigenvalues and eigenvectors to understand the data structure.

# Simulated data
df <- data.frame(
glucose = rnorm(100, 100, 15),
cholesterol = rnorm(100, 200, 25),
bp = rnorm(100, 120, 10)
)

# Correlation
cor(df, method = "pearson")
cor(df, method = "spearman")

# Eigen decomposition
e <- eigen(cor(df))
e$values
e$vectors

9. Analyze the Heart Disease dataset by exploring its structure, target variable, and
missing values. Visualize correlations, split data, and train a Logistic Regression
model. Evaluate using accuracy, precision, recall, and F1-score. Apply PCA for
dimensionality reduction and assess if performance improves for better heart
disease prediction.

heart <- read.csv("heart.csv")

# Train-test split
set.seed(123)
train_idx <- sample(1:nrow(heart), 0.7 * nrow(heart))
train <- heart[train_idx, ]
test <- heart[-train_idx, ]

# Logistic regression
model <- glm(target ~ ., data=train, family="binomial")
pred <- predict(model, newdata=test, type="response")
pred_class <- ifelse(pred > 0.5, 1, 0)

# Evaluation
 conf_matrix <- table(Predicted=pred_class, Actual=test$target)
accuracy <- sum(diag(conf_matrix)) / sum(conf_matrix)
accuracy

# PCA
pca <- prcomp(train[,-ncol(train)], scale. = TRUE)
summary(pca)

10. A real estate company aims to predict house prices based on factors such as square
footage, number of bedrooms, and location. Formulate the problem as a multiple
linear regression equation and express it in matrix form. Using the normal
equation, calculate the regression coefficients to fit the model. Discuss how
multicollinearity among predictor variables affects the reliability of the regression
model and explain how Principal Component Regression (PCR) can be used to
address multicollinearity issues while improving predictive accuracy.

# Sample data
data <- data.frame(
sqft = c(1000, 1500, 2000),
bedrooms = c(2, 3, 4),
location = c(1, 2, 3), # encoded location
price = c(200000, 300000, 400000)
)

# Design matrix X (with bias term)

X <- as.matrix(cbind(1, data$sqft, data$bedrooms, data$location))
y <- as.matrix(data$price)

# Normal equation
beta <- solve(t(X) %*% X) %*% t(X) %*% y
beta

# PCR to fix multicollinearity

library(pls)
model <- pcr(price ~ ., data=data, scale=TRUE, validation="CV")
summary(model)