U18AII5202 – EXPLORATORY DATA
ANALYSIS AND VISUALIZATION
UNIT I – ANALYSIS TECHNIQUES
TOPIC: INTRODUCTION TO STATISTICAL
LEARNING AND R-PROGRAMMING
Matrix
Example 1: To modify a matrix
# create 2 by 2 matrix
matrix1 <- matrix(c(1, 2, 3, 4), nrow = 2, ncol = 2)
# print original matrix
print(matrix1)
# change value at 1st row, 2nd column to 5
matrix1[1,2] = 5
# print updated matrix
print(matrix1)
Matrix
Example 2: To combine two matrices
# create two 2 by 2 matrices
even_numbers <- matrix(c(2, 4, 6, 8), nrow = 2, ncol =
2)
odd_numbers <- matrix(c(1, 3, 5, 7), nrow = 2, ncol = 2)
# combine two matrices by column
total1 <- cbind(even_numbers, odd_numbers)
print(total1)
# combine two matrices by row
total2 <- rbind(even_numbers, odd_numbers)
print(total2)
Matrix
Example 3: Check if Element Exists in R Matrix
matrix1 <- matrix(c("Sabby", "Cathy", "Larry", "Harry"),
nrow = 2, ncol = 2)
"Larry" %in% matrix1 # TRUE
"Kinsley" %in% matrix1 # FALSE
List
Example 1: Access list elements
list1 <- list(24, "Sabby", 5.4, "Nepal")
# access 1st item
print(list1[1]) # 24
# access 4th item
print(list1[4]) # Nepal
List
Example 2: Add/remove list
elements
list1 <- list(24, "Sabby", 5.4,
"Nepal")
# change element at index 2
list1[2] <- "Cathy"
# print updated list
print(list1)
# using append() function
append(list1, 3.14)
# remove 4th item
print(list1[-4])
List
Example 3: Check list elements
list1 <- list(24, "Sabby", 5.4, "Nepal")
"Sabby" %in% list1 # TRUE
"Kinsley" %in% list1 # FALSE
List
Example 4: Create list using vectors
empId = c(1, 2, 3, 4)
empName = c("Debi", "Sandeep", "Subham", "Shiba")
numberOfEmp = 4
empList = list(empId, empName, numberOfEmp)
print(empList)
Example 5: Creating a named list
my_named_list <- list(name = "Sudheer", age = 25, city
= "Delhi")
# Printing the named list
print(my_named_list)
List
Example 6: Accessing R List Components
empId = c(1, 2, 3, 4)
empName = c("Debi", "Sandeep", "Subham", "Shiba")
numberOfEmp = 4
empList = list(
"ID" = empId,
"Names" = empName,
"Total Staff" = numberOfEmp
)
print(empList)
# Accessing components by names
cat("Accessing name components using $ command\n")
print(empList$Names)
List
Example 7: Accessing R List Components
empId = c(1, 2, 3, 4)
empName = c("Debi", "Sandeep", "Subham", "Shiba")
numberOfEmp = 4
empList = list(
"ID" = empId,
"Names" = empName,
"Total Staff" = numberOfEmp
)
print(empList)
print(empList[[2]])
print(empList[[2]][2])
print(empList[[1]][4])
List
Example 8: Modifying Components of a List
# Creating a list by naming all its components
empId = c(1, 2, 3, 4)
empName = c("Debi", "Sandeep", "Subham", "Shiba")
numberOfEmp = 4
empList = list(
"ID" = empId,
"Names" = empName,
"Total Staff" = numberOfEmp
)
empList$`Total Staff` = 5
empList[[1]][5] = 5
empList[[2]][5] = "Kamala"
print(empList)
List
Example 9: Concatenation of lists Example 10: Converting List to
Vector
# Create two lists.
lst1 <- list(1,2,3) # Create lists.
lst2 <- list("Sun","Mon","Tue") lst <- list(1:5)
print(lst)
# Merge the two lists.
new_list <- c(lst1,lst2) # Convert the lists to vectors.
vec <- unlist(lst)
# Print the merged list.
print(new_list) print(vec)
List
Example 11: Converting list to matrix
# Defining list
lst1 <- list(list(1, 2, 3),
list(4, 5, 6))
# Convert list to matrix
mat <- matrix(unlist(lst1), nrow = 2, byrow =
TRUE)
# Print matrix
cat("\nAfter conversion to matrix:\n")
print(mat)
cat("Class:", class(mat), "\n")
Dataframes
Example 1: Creating a dataframe
# creating a data frame
friend.data <- data.frame(
friend_id = c(1:5),
friend_name = c("Sachin", "Sourav",
"Dravid", "Sehwag",
"Dhoni"),
stringsAsFactors = FALSE
)
# print the data frame
print(friend.data)
Dataframes
Example 2: Structure of a dataframe
friend.data <- data.frame(
friend_id = c(1:5),
friend_name = c("Sachin", "Sourav",
"Dravid", "Sehwag",
"Dhoni"),
stringsAsFactors = FALSE
)
# using str()
print(str(friend.data))
Dataframes
Example 3: Summary of a dataframe
friend.data <- data.frame(
friend_id = c(1:5),
friend_name = c("Sachin", "Sourav",
"Dravid", "Sehwag",
"Dhoni"),
stringsAsFactors = FALSE
)
# using summary()
print(summary(friend.data))
Dataframes
Example 4: Extracting data from a dataframe
friend.data <- data.frame(
friend_id = c(1:5),
friend_name = c("Sachin", "Sourav",
"Dravid", "Sehwag",
"Dhoni"),
stringsAsFactors = FALSE
)
# Extracting friend_name column
result <- data.frame(friend.data$friend_name)
print(result)
Dataframes
Example 5: Expanding data from a dataframe
friend.data <- data.frame(
friend_id = c(1:5),
friend_name = c("Sachin", "Sourav",
"Dravid", "Sehwag",
"Dhoni"),
stringsAsFactors = FALSE
)
# Expanding data frame
friend.data$location <- c("Kolkata", "Delhi",
"Bangalore", "Hyderabad",
"Chennai")
resultant <- friend.data
# print the modified data frame
print(resultant)
Dataframes
Example 6: Accessing elements from a dataframe
friend.data <- data.frame(
friend_id = c(1:5),
friend_name = c("Sachin", "Sourav",
"Dravid", "Sehwag",
"Dhoni"),
stringsAsFactors = FALSE
)
# Access Items using []
friend.data[1]
# Access Items using [[]]
friend.data[['friend_name']]
# Access Items using $
friend.data$friend_id
Dataframes
Example 7: Number of rows and columns in a
dataframe
friend.data <- data.frame(
friend_id = c(1:5),
friend_name = c("Sachin", "Sourav",
"Dravid", "Sehwag",
"Dhoni"),
stringsAsFactors = FALSE
)
# find out the number of rows and clumns
dim(friend.data)
Dataframes
Example 8: Add Rows in R Data Frame
# Creating a dataframe representing products in a
store
Products <- data.frame(
Product_ID = c(101, 102, 103),
Product_Name = c("T-Shirt", "Jeans", "Shoes"),
Price = c(15.99, 29.99, 49.99),
Stock = c(50, 30, 25)
)
New_Product <- c(104, "Sunglasses", 39.99, 40)
Products <- rbind(Products, New_Product)
print(Products)
Dataframes
Example 9: Add Columns in R Data Frame
# Existing dataframe representing products in a store
Products <- data.frame(
Product_ID = c(101, 102, 103),
Product_Name = c("T-Shirt", "Jeans", "Shoes"),
Price = c(15.99, 29.99, 49.99),
Stock = c(50, 30, 25)
)
Discount <- c(5, 10, 8)
Products <- cbind(Products, Discount)
# Rename the added column
colnames(Products)[ncol(Products)] <- "Discount"
print(Products)
Dataframes
Example 10: Remove Rows in R Data Frame
library(dplyr)
# Create a data frame
data <- data.frame(
friend_id = c(1, 2, 3, 4, 5),
friend_name = c("Sachin", "Sourav", "Dravid",
"Sehwag", "Dhoni"),
location = c("Kolkata", "Delhi", "Bangalore",
"Hyderabad", "Chennai")
)
# Remove a row with friend_id = 3
data <- subset(data, friend_id != 3)
data
Dataframes
Example 11: Remove Columns in R Data Frame
library(dplyr)
# Create a data frame
data <- data.frame(
friend_id = c(1, 2, 3, 4, 5),
friend_name = c("Sachin", "Sourav", "Dravid",
"Sehwag", "Dhoni"),
location = c("Kolkata", "Delhi", "Bangalore",
"Hyderabad", "Chennai")
)
data
# Remove the 'location' column
data <- select(data, -location)
data
Dataframes
Example 12: Combine Data frame Vertically
# Creating two sample dataframes
df1 <- data.frame(
Name = c("Alice", "Bob"),
Age = c(25, 30),
Score = c(80, 75)
)
df2 <- data.frame(
Name = c("Charlie", "David"),
Age = c(28, 35),
Score = c(90, 85)
)
# Combining the dataframes using rbind()
combined_df <- rbind(df1, df2)
print(combined_df)
Dataframes
Example 13: Combine Data frame Horizontally
df1 <- data.frame(
Name = c("Alice", "Bob"),
Age = c(25, 30),
Score = c(80, 75)
)
df2 <- data.frame(
Height = c(160, 175),
Weight = c(55, 70)
)
# Combining the dataframes using cbind()
combined_df <- cbind(df1, df2)
print(combined_df)
