1

STATISTICAL COMPUTING AND R PROGRAMMING

MODULE-05
Simple linear regression, multiple linear regression, linear model selection and diagnostic s.
Advanced graphics: plot customization, plotting regions and margins, point and click
coordinate interaction, customizing traditional R plots, specialized text and label
notation.Defining colors and plotting in higher dimensions, representing and using color, 3D
scatter plots.

Regression: The linear regression is defined as the analyze the relationship between the
independent and dependent variable is called regression.
In statistics, regression refers to a statistical method used to examine the relationship between
one or more independent variables and a dependent variable.
It aims to model and understand how changes in the independent variables are associated with
changes in the dependent variable.
Regression analysis helps in:
Predicting or estimating the value of the dependent variable based on the values of one
or more independent variables.
Understanding the strength and nature of the relationship between variables.
Making inferences about the population based on sample data.
There are various types of regression models,
1. Simple Linear Regression: Examining the relationship between two variables,
typically one independent variable and one dependent variable.
2. Multiple Linear Regression: Exploring the relationship between a dependent
variable and multiple independent variables.
3. Logistic Regression: Used when the dependent variable is categorical, predicting the
probability of an event occurring.
NOTE:
1) Regression analysis involves fitting a regression model to the data, assessing the
model's goodness of fit, evaluating the significance of variables, and using the model
for prediction or inference.
2) It's widely used across numerous fields to understand relationships, make predictions,
and inform decision-making based on data.
3) The dependent variable(Y):It is the outcome variable or response variable. That want
to be predict.

JSSCACS 2023-24 ROOPA

 2
STATISTICAL COMPUTING AND R PROGRAMMING

4) The Independent variable(X):The independent variable is predictor variable that

influence the dependent variable.
An Example of a Linear Relationship:
Let's consider an example illustrating a linear relationship between two variables, such as the
relationship between a person's age and their reaction time.
Variables:
Dependent Variable (Y): Reaction time (in milliseconds) response variable - this is what we
want to predict or understand.
Independent Variable (X): Age of the person (in years) - this is the predictor variable.
Assumption:
We assume that younger individuals tend to have faster reaction times, and as people age, their
reaction times might increase.
Data Collection:
Gather data on reaction times and ages from individuals across different age groups.
Analysis:
After collecting the data, plot the reaction times against ages on a scatter plot.
If a linear relationship exists, the points might tend to form a pattern where, on average, as age
increases, reaction time also tends to increase or decrease in a linear manner.
Linear Relationship:
A positive linear relationship would imply that as age increases, reaction time increases.
A negative linear relationship would suggest that as age increases, reaction time decreases.
Modeling:
Use linear regression to fit a line to the data and quantify the relationship between age and
reaction time.
By studying this relationship, we can determine if there's a correlation between age and
reaction time, and if so, whether it's linear, helping us understand how one variable might
affect the other. This example demonstrates how linear relationships between variables can be
explored and modelled in statistics.
Simple Linear Regression:
Definition: Simple linear regression is a statistical method used to model the relations hip
between two continuous variables.
OR
A Simple Linear regression predict a one dependent and one independent variable.

JSSCACS 2023-24 ROOPA

 3
STATISTICAL COMPUTING AND R PROGRAMMING

It assumes that there is a linear relationship between the predictor variable (often denoted as
X) and the outcome variable (often denoted as Y).
The simple linear regression model can be represented as:
Y=β0+β1⋅X+ε
Where:
 Y is the dependent variable (outcome or response variable).
 X is the independent variable (predictor or explanatory variable).
 β0 is the intercept (the value of Y when X is 0).
 β1 is the slope (the change in Y for a one-unit change in X).
 ε represents the error term (the difference between the predicted and actual values).
The goal of simple linear regression is to estimate the values of β0 and β1 that minimize the
sum of squared differences between the observed Y values and the values predicted by the
model for given X values.
Applications:
Simple linear regression is a foundational statistical technique used to establish relationship s
between two quantitative variables.
1.Economics and Finance: In finance, it's used to model the relationship between variables
like interest rates and stock prices or GDP and unemployment rates. Economists might employ
it to analyze the effect of inflation on consumer spending.
2.Market Research: Simple linear regression helps in analyzing the impact of advertising
expenditure on sales. It assists in understanding how changes in advertising spending might
affect product sales.
3.Healthcare and Medicine: Medical researchers often use simple linear regression to study
the relationship between a drug dosage and its effectiveness or to examine the impact of certain
lifestyle factors (like diet, exercise) on health indicators.
4.Education: Simple linear regression can be applied in educational research to investiga te
the relationship between study hours and exam scores or to predict student performance based
on various factors.
5. Sports Analytics: Analysts might use it to study the relationship between the number of hours of
practice and athletic performance or to predict team performance based on player statistics.

JSSCACS 2023-24 ROOPA

 4
STATISTICAL COMPUTING AND R PROGRAMMING

SIMPLE LINEAR REGRESSION FEATURES IN STATISTICS

1. One Dependent Variable: It involves predicting or explaining the behavior of a single
dependent variable based on changes in one independent variable.
2. Straight-line Relationship: The relationship between the independent and dependent
variables is assumed to be linear. The regression equation is represented by a straight line: Y
= β₀ + β₁X + ε, where Y is the dependent variable, X is the independent variable, β₀ is the
intercept, β₁ is the slope, and ε is the error term.
3. Minimization of Residuals: The regression line is determined by minimizing the sum of
squared differences (residuals) between the observed values and the values predicted by the
model.
4. Statistical Inference: It allows for statistical hypothesis testing to determine the significance
of the relationship between variables. This includes testing whether the slope of the regression
line is significantly different from zero and assessing the overall goodness-of-fit of the model.
5. Assumptions: Simple linear regression assumes that there is a linear relationship between the
variables, residuals are normally distributed, residuals have constant variance
(homoscedasticity), and there is independence among observations.
6. Interpretation of Coefficients: The slope coefficient (β₁) represents the change in the
dependent variable for a one-unit change in the independent variable. The intercept (β₀)
represents the value of the dependent variable when the independent variable is zero (if such
an interpretation is meaningful in the context).
7. Prediction: It allows for predicting the value of the dependent variable based on a given value
of the independent variable, within the range of the observed data.

The goal of simple linear regression is to estimate the values of β0 and β1 that minimize the
sum of squared differences between the observed Y values and the values predicted by the
model for given X values.
Y== β₀ + β₁X + ε

The formula to estimate the slope (β1) is: The intercept(𝛽0)

𝑛
(𝑥𝑖 −𝑥̅) (𝑦𝑖 −𝑦̅ )
𝛽1 = ∑ ( 𝑥𝑖 −𝑥̅)2
𝛽0 = 𝑌̅ − 𝛽0 . 𝑋̅
𝑖̇=1
And once the slope (β1) is calculated, the intercept (β0) can be found using the formula:
Where:
n is the number of data points.
ˉXˉ and ˉYˉ are the means of X and Y respectively.

JSSCACS 2023-24 ROOPA

 5
STATISTICAL COMPUTING AND R PROGRAMMING

Estimating the Intercept and Slope Parameters

* The goal is to use your data to estimate the regression parameters, yield in g
the estimates β̂0 and β̂1 . this is referred to as fitting the linear model.
* In this case, the data comprise n pairs of observations for each individ ua l.
The fit- ted model of interest concerns the mean response value, denoted
yˆ, for a specific value of the predictor, x, and is written as follows:

yˆ = βˆ0 + βˆ1. x
oR
E[Y ] = βˆ0 + βˆ1 .x

OR
E[Y|X = x] = βˆ0 + βˆ1 .x

Alternative notation such as E[Y ] or E[Y|X = x] is used on the left side of to

emphasize the fact that the model gives the mean (that is, the expected value)
of the response. For compactness, many simply use something like yˆ, as shown
here.
Let your n observed data pairs be denoted x i and yi for the predictor and
response variables, respectively; i = 1, . . . , n. Then, the parameter esti- mates
for the simple linear regression function are

β̂1 = 𝛒𝐱𝐲𝐒𝐘 and 𝛽0 = 𝑌̅ − 𝛽1 . 𝑋̅

𝐬𝐗

𝑛
(𝑥𝑖 − 𝑥̅ )(𝑦𝑖 − 𝑦̅ )
𝛽1 = ∑
(𝑥𝑖 − 𝑥̅ )2
𝑖̇ =1

Problems:
X={2,3,5,7,9} y={4,5,7,10,15}
Step1:
∑𝑥 26
To calculate mean 𝑋̅ = = =5.2
𝑛 5
∑𝑌 41
𝑌̅ = = =8.2
𝑛 5

Step2:To calculate Slope 𝛽1

𝑛
( 𝑥𝑖 −𝑥̅ ) (𝑦𝑖 −𝑦̅ ) 3.4
𝛽1 = ∑ ( 𝑥𝑖 −𝑥̅ ) 2
= =0.098
𝑖̇=1 34.64̇

JSSCACS 2023-24 ROOPA

 6
STATISTICAL COMPUTING AND R PROGRAMMING

1.(𝑥𝑖 − 𝑥̅ )(𝑦𝑖 − 𝑦̅)=(2-5.2)(4-8.2)+(3-5.2)(5-8.2)+……=3.4

2. (𝑥𝑖 − 𝑥̅ )2=((2-5.2)^2)+((3-5.2)^2)+…=34.64
Step3:To calculate Intercept (𝛽0)
𝛽0 = 𝑌̅ − 𝛽1 . 𝑋̅
=8.2-0.098*5.2
=7.69
Step4:Write the linear Equation.
To substitute the value 𝛽0 and 𝛽1 into the equation
yˆ = βˆ0 + βˆ1. x
Y=7.69+0.098x

To Write in R program
The syntax lm function:

lm(formula, data)

S.No Parameters Description

1. Formula It is a symbol that presents the relationship between x and y.

2. Data It is a vector on which we will apply the formula.

Example:
#Creating input vector for lm() function
x <- c(2,3,5,7,9)
y <- c(4,5,7,10,15)

# Applying the lm() function.

relationship_model<- lm(y~x)

#Printing the coefficient

print(relationship_model)

JSSCACS 2023-24 ROOPA

 7
STATISTICAL COMPUTING AND R PROGRAMMING

Output:
Call:
lm(formula = y ~ x)

Coefficients:
(Intercept) slope x
7.67 0.098

The predict() Function

Now, we will predict the weight of new persons with the help of the predict() function. There
is the following syntax of predict function:
Syntax:
predict(object, newdata)

S.No Parameter Description

1. object It is the formula that we have already created using the lm() function.

2. Newdata It is the vector that contains the new value for the predictor variable.

Source Code:
#Creating input vector for lm() function
x <- c(2,3,5,7,9)
y <- c(4,5,7,10,15)

# Applying the lm() function.

relationship_model<- lm(y~x)

# Finding the weight of a person with height 9.

z <- data.frame(x = 9)
predict_result<- predict(relationship_model,z)
print(predict_result)
Output:
1
13.96951

JSSCACS 2023-24 ROOPA

 8
STATISTICAL COMPUTING AND R PROGRAMMING

Plotting Regression

Now, we plot out prediction results with the help of the plot() function. This function takes
parameter x and y as an input vector and many more arguments.
plot(y,x,col = "red",main = "Height and Weight Regression",abline(lm(x~y)),cex = 1.3,pch =
16,xlab = "Weight in Kg",ylab = "Height in cm")
# Saving the
file. dev.off()

Multiple linear regression:

The Multiple linear regression are the Multiple relationship between independent and
dependent two or more variable.

OR

A Multiple Linear regression predict a one dependent and Tow-more independent variable.

It is the extension of the simple linear regression. The multiple linear regression model
is used to determine a mathematical relationship among several random variable.
Multilinear regression, also known as multiple linear regression, is a statistica l method
used to analyze the relationship between multiple independent variables and a
dependent variable.

JSSCACS 2023-24 ROOPA

 9
STATISTICAL COMPUTING AND R PROGRAMMING

It extends simple linear regression, which involves one independent variable, to

incorporate several predictors.
The goal of multiple linear regression is to create a model that best fits the relations hip
between these variables, allowing prediction or understanding of how changes in the
independent variables affects the dependent variable.

EXAMPLE:

let's consider a real-life example of multiple linear regression. Imagine you're trying to
predict house prices based on various factors such as square footage, number of bedrooms,
and distance from the city center.

Solution: In the case of multiple linear regression for predicting house prices:

Dependent Variable: House price.

Independent Variables (Predictors):

1. Square Footage: The size of the house in square feet.
2. Number of Bedrooms: The count of bedrooms in the house.
3. Distance from City Center: How far the house is from the city center.

Using multiple linear regression, a model can be created based on historical data of houses
sold. This model will estimate the house price by considering all these factors
simultaneously. For instance, the model might suggest that for every additional square foot,
the price increases by a certain amount, or that being closer to the city center correlates with
higher prices. This information helps in making predictions about house prices for new
properties based on their features.

Applications/Uses of Multiple Linear Regression:

Multiple linear regression finds applications across various fields:

1. Economics: Predicting factors influencing GDP growth, inflation rates, or
unemployment based on multiple economic indicators.
2. Marketing: Determining how various marketing channels, pricing strategies, and
demographics affect sales figures.
3. Healthcare: Analyzing how factors like age, lifestyle, and medical history impact the
prevalence of certain diseases or predicting patient outcomes based on multiple health
parameters.
4. Finance: Predicting stock prices or asset values based on various financial indicators

JSSCACS 2023-24 ROOPA

 10
STATISTICAL COMPUTING AND R PROGRAMMING

such as interest rates, market indices, and company performance metrics.

5. Environmental Science: Predicting air or water quality based on factors like pollutio n
levels, weather conditions, and geographical variables.

Extending the simple model to a multiple model

To determine the value of a continuous response variable Y given the values of
p > 1 independent explanatory variables X1 , X2 , . . ., X p . The overarching model
is defined as
Y = β0 + β1 X1 + β2 X2 + . . . + βp Xp + ϵ
Here:
 Y is the dependent variable
 X1,X2,X3 are the independent variables (square footage, number of
bedrooms, distance from the city center).
 β0 is the intercept (constant term).
 β1,β2,β3 are the coefficients that represent the impact of each independent
variable on the dependent variable.
 ϵ represents the error term.
Example, the estimated equation might look like:
House Price=β0+β1×Square Footage+β2×Number of Bedrooms+β3×Distance from
City Center
Each coefficient (β) signifies the change in the house price for a unit change in the
respective independent variable, assuming the other variables remain constant.
NOTE:
 where β0 , . . . , βp are the regression coefficients and, as before, you assume
independent, normally distributed residuals∼ € ~ N(0, σ) around the mean.
 In practice, you have n data records; each record provides values for
{
each of the predictors X j ; j = { 1, . . ., p } .
 The model to be fitted is }given in terms of the mean respo ns e,
conditional upon a particular realization of the set of explana to r y
variables

ŷ = E[Y | X1 = x 1 , X2 = x 2 , . . . , X p = x p ] = βˆ0 + βˆ1 x 1 + βˆ2 x 2 + . . . + βˆp x p ,

where the β̂ j s represe nt estimates of the regression coefficients.

JSSCACS 2023-24 ROOPA

 11
STATISTICAL COMPUTING AND R PROGRAMMING

Estimating in Matrix Form

The computations involved in minimizing this squared distance are made much easier
by a matrix representation of the data. When dealing with n multivariate observations

Y= X.β+€

where Y and € denote n × 1 column matrices such the

𝒚𝟏 €𝟏 𝜷𝟎
𝒚𝟐 €𝟐 𝜷𝟏
𝒚= : €= : 𝜷= :
: : :
[𝒚𝒏] [€𝒏] [𝜷𝒏]

1 𝑋1,1 𝑋𝑝,1
x = [1 𝑋1,2 𝑋𝑝,2 ]
1 𝑋1,𝑛 𝑋𝑝,𝑛

=(𝑥 𝑇 ∗ 𝑋) −1𝑥 𝑇 ∗ 𝑌

NOTE:
T
* The symbol represents matrix multiplication, the superscript rep- resents the
transpose, and −1 represents the inverse when applied to matrice s
* Extending the size of β and X (note the leading column of 1s in X) to
create structures of size p + 1 (as opposed to just the number of
predictors p) allows for the estimation of the overall intercept β0.

Terminology:
1. lurking variable : A lurking variable influences the response, anothe r
predictor, or both, but goes unmeasured (or is not included) in a
predictive model
2. nuisance or extraneous variable :A nuisance or extraneous variable is a
predictor of secondary or no interest that has the potential to confo un d
relationships between other variables and so affect your estimates of the
other regression coefficients.
3. Multiple Linear Regression: A statistical method to model the relations hip
between multiple independent variables and a dependent variable.

JSSCACS 2023-24 ROOPA

 12
STATISTICAL COMPUTING AND R PROGRAMMING

4. Dependent Variable: Also known as the response variable, it's the variable
being predicted or explained by the independent variables in the model.
5. Independent Variables/Predictors: Variables used to predict or explain the
variation in the dependent variable.
6. Coefficients: These represent the weights or slopes assigned to each
independent variable, indicating the strength and direction of their influence on
the dependent variable. In the regression equation Y=β0+β1X1+β2X 2+...+β n

Xn+ϵ , coefficients ( β0,β1,β2,...,βn ) represent the slopes of the independent

variables. These coefficients quantify the relationship between each
independent variable and the dependent variable.
7. Intercept: The constant term in the regression equation that accounts for the
value of the dependent variable when all independent variables are zero.
8. Residuals: The differences between the observed values of the dependent
variable and the values predicted by the regression model.
9. Multicollinearity: The phenomenon where independent variables in a
regression model are highly correlated, which can affect the model's stability
and interpretability.
10. Adjusted R-squared: A metric that represents the proportion of variation in
the dependent variable explained by the independent variables, adjusted for the
number of predictors in the model.
A step-by-step to performing multiple linear regression using statistical methods:
Step 1: Formulate the Hypothesis
Define the research question and hypothesize the relationship between the dependent variable
and multiple independent variables.
Step 2: Data Collection and Cleaning
collect relevant data for the dependent and independent variables. Clean the data by handling
missing values, outliers, and ensuring the data types are appropriate.
Step 3: Explore the Data
Perform exploratory data analysis (EDA) to understand distributions, correlations, and
relationships between variables using techniques like scatter plots, correlation matrices, etc.
Step 4: Model Specification
Formulate the multiple linear regression model:
Y=β0+β1X1+β2X2+…+βnXn+ϵ

JSSCACS 2023-24 ROOPA

 13
STATISTICAL COMPUTING AND R PROGRAMMING

Where:
Y is the dependent variable.
X1,X2,…,Xn are the independent variables.
β0,β1,…,βn are the coefficients.
ϵ is the error term.
Step 5: Estimate Coefficients
Use statistical methods (such as the least squares method) to estimate the coefficients that
minimize the sum of squared differences between predicted and actual values.
Step 6: Assess Model Fit
Evaluate the goodness of fit of the model by examining metrics like R-squared, adjusted R-
squared, p-values of coefficients, and residuals analysis.
Step 7: Interpret Results
Interpret the coefficients to understand the relationship between the independent variables
and the dependent variable.
Step 8: Validate the Model
Validate the model's performance on a separate dataset or using cross-validation technique s
to ensure its reliability and generalizability.
Step 9: Make Predictions
Use the validated model to make predictions on new or unseen data based on the established
relationships.

SYNTAX FOR MULPLE LINEAR REGRESSION IN R

lm(formula,data)

S.No Parameters Description

1. Formula It is a symbol that presents the relationship between x and y.

2. Data It is a vector on which we will apply the formula.

Multiple linear regression in R using statistical methods step by step.

Step 1: Load the Required Libraries :

Before you begin, make sure to load the necessary R libraries. For linear regression, you'll
typically use the lm() function, which is in the base R package

# Load required libraries

library(stats) # for basic statistics functions

JSSCACS 2023-24 ROOPA

 14
STATISTICAL COMPUTING AND R PROGRAMMING

Step 2: Prepare Your Data:

A dataset with the variables you want to use for the regression analysis. Ensure your data is in
a format that R can work with, like a dataframe.
Let's assume you have a dataframe named data with your variables: dependent_var ,
independent_var1 , independent_var2 , etc.

Step 3: Fit the Multiple Linear Regression Model

Use the lm() function to fit the multiple linear regression model. The basic syntax for the lm()
function is:
# Fit the multiple linear regression model
model <- lm(dependent_var ~ independent_var1 + independent_var2 + ..., data = your_data)
Replace dependent_var , independent_var1 , independent_var2 , etc., with the actual
column names from your dataset.
Step 4: Analyze the Model
Once you've fitted the model, you can examine various aspects of it using different functions:
# print of the regression model
print(model)
Example:
# Load required libraries
library(stats)
# Creating a dataframe (replace this with your dataset)
data <- data.frame(
dependent_var = c(10, 15, 20, 25, 30),
independent_var1 = c(3, 6, 7, 9, 11),
independent_var2 = c(5, 8, 11, 14, 17)
)
# Fit the multiple linear regression model
model <- lm(dependent_var ~ independent_var1 + independent_var2, data = data)
# Summary of the regression model
print(model)
OUTPUT:Call:
lm(formula = dependent_var ~ independent_var1 + independent_var2, data = data)
Coefficients:
(Intercept) independent_var1 independent_var2

JSSCACS 2023-24 ROOPA

 15
STATISTICAL COMPUTING AND R PROGRAMMING

1.667 0.000 1.667

This code creates a dataset and performs multiple linear regression using lm(), where Y is
regressed on X1, X2, and X3. The summary() function provides detailed information about
the regression results, including coefficients, p-values, R-squared, etc.
Replace Y, X1, X2, X3, and the sample dataset with your actual data and variable names to
apply multiple linear regression to your specific case in R.
Difference between Simple linear Regression and Multiple linear Regression
Simple linear Regression Multiple linear Regression

1. A Simple Linear regression 1. A Multiple Linear regression predict a one

predict a one dependent and one dependent and Tow-more independent variable.
independent variable.

2.Y = β0 + β1 X1 + β2 + ϵ 2. Y = β0 + β1 X1 + β2 X2 + . . . + βp Xp
+ϵ

Write a program to create an any application of Linear Regression in multivariate

context forpredictive purpose.

Source code:
# Creating sample data
set.seed(42)
bedrooms <- sample(1:5, 100, replace = TRUE)
house_size <- rnorm(100, mean = 1500, sd = 300)
house_price <- 50000 + 20000 * bedrooms + 100 * house_size + rnorm(100, mean = 0, sd =
5000)

data <- data.frame(bedrooms, house_size, house_price)

# Fitting a multiple linear model

model <- lm(house_price ~ bedrooms + house_size, data = data)

# Creating new data for prediction

JSSCACS 2023-24 ROOPA

 16
STATISTICAL COMPUTING AND R PROGRAMMING

new_data <- data.frame(bedrooms = c(3, 4, 2), house_size = c(1600, 1800, 2000))

# Predicting house_price using the model

new_predictions <- predict(model, newdata = new_data)
new_predictions

OUTPUT:
1 2 3
127690.4 147650.5 107250.9
 Sample data for bedrooms , house_size , and house_price is created.
 A multiple linear regression model is fitted (model) using lm() with house_price as
the dependent variable and bedrooms and house_size as independent variables.
 New data (new_data) with values for bedrooms and house_size for three houses is
created for which we want to predict house_price .
 The predict() function is used to predict house_price for the new data based on the
model.
 The output provides predicted house prices for the three houses specified in the
new_data.

# Creating sample data

set.seed(42)
bedrooms <- sample(1:5, 100, replace = TRUE)
house_size <- rnorm(100, mean = 1500, sd = 300)
house_price <- 50000 + 20000 * bedrooms + 100 * house_size + rnorm(100, mean = 0, sd =
5000)

data <- data.frame(bedrooms, house_size, house_price)

# Fitting a multiple linear model

model <- lm(house_price ~ bedrooms + house_size, data = data)

JSSCACS 2023-24 ROOPA

 17
STATISTICAL COMPUTING AND R PROGRAMMING

# Finding confidence intervals for coefficients

conf_intervals <- confint(model)
conf_intervals
OUTPUT:
2.5 % 97.5 %
(Intercept) 48610.926 51343.900
bedrooms 18251.833 21500.128
house_size 82.262 115.796
 Generates sample data for bedrooms , house_size , and house_price
 Fits a multiple linear regression model ( model ) using lm() with house_price as the
dependent variable and bedrooms and house_size as independent variables.
 Uses the confint() function to compute confidence intervals for the coefficients in the

fitted model.
 The conf_intervals output will provide the lower and upper bounds of the confidence

intervals for each coefficient in the multiple linear regression model.

 Each row corresponds to a coefficient in the regression model.
 Columns '2.5 %' and '97.5 %' represent the lower and upper bounds of the 95%
confidence interval for each coefficient, respectively.
 For instance, the confidence interval for the bedrooms coefficient ranges from

approximately 18251.833 to 21500.128. This interval provides a range within which

we can be reasonably confident that the true value of the coefficient lies.
LINEAR MODEL SELECTION AND DIAGNOSTICS

 Linear model selection and diagnostics involve choosing the appropriate linear model
for a given dataset and assessing its performance.
 Linear model selection methods are a critical part of statistical analysis with a wide
range of applications, particularly in high-dimensional data analysis, where the
number of variables is much larger than the sample size.
 In linear model selection are the select the best model based on the criteria.
 The Linear selection model is also called as the statistical model
The linear selection models are selected or balancing few factors.

JSSCACS 2023-24 ROOPA

 18
STATISTICAL COMPUTING AND R PROGRAMMING

1. Goodness-of-fit
2. Complexity
1. Goodness-of-fit: Goodness-of- fit refers to the goal of obtaining a model that best
represents the relationships between the response and the predictor (or pre-
dictors).
2. Complexity: Complexity describes how complicated a model is; this is always tied
to the number of terms in the model that require estimation—the inclusion of
more predictors and additional functions (such as polynomial transforma tions
and interactions) leads to a more complex model.
OR
To represents the complicated model selection from various theoretical and practical
challenges
Principle of Parsimony:
Statisticians refer to the balancing act between goodness-of- fit and complexity as the
principle of parsimony,
where the goal of the associated model selection is to find a model that’s as simple as possib le
(in other words, with relatively low complexity), without sacrificing too much goodness-
of-fit.
LINEAR MODEL SELECTION:
 In statistics, several model selection algorithms help in choosing the best-fitting model among
a set of candidate models.
 A model selection algorithm is to sift through your available explana to r y
variables
1. Forward Selection: Begins with an empty model and iteratively adds predictors that
most improve the model fit until a stopping criterion is met.
2. Backward Elimination: Starts with a model containing all predictors and removes the
least significant ones iteratively until a stopping criterion is reached.
3. Stepwise Selection:Combines forward and backward selection methods, allowing both
addition and removal of predictors based on certain criteria.
4. Best Subset Selection: Fits all possible combinations of predictors and selects the
model that best fits the data based on a chosen criterion (e.g., AIC, BIC).

The Partial F-Test:

The partial F-test is probably the most direct way to compare several differ- ent models. It
looks at two or more nested models, where the smaller, less complex model is a reduced

JSSCACS 2023-24 ROOPA

 19
STATISTICAL COMPUTING AND R PROGRAMMING

version of the bigger, more complex model. Formally, let’s say you’ve fitted two linea r
regressions models as follows:

ŷ redu = β̂0 + β̂1 x 1 + β̂2 x 2 + . . . + β̂ p x p

ŷ full = β̂0 + β̂1 x 1 + β̂2 x 2 + . . . + β̂ p x p + . . . + β̂ q x q

Here, the reduced model, predicting ŷ redu, has p predictors, plus one intercept. The
full model, predicting ŷ full , has q predictor terms.
provide a statistically significant improvement in goodness-of- fit. The partial F-
test addresses these hypotheses: −

H0 : βp+1 = βp+2 = . . . = βq = 0
HA : At least one of the βj ≠ 0 (for j = p, . . . , q)
ForwardSelection:
Forward selection is a model selection technique in linear regression where predictors are
incrementally added to the model based on statistical criteria until no further improvement is
observed
Example:
# Sample data
set.seed(123)
data <- data.frame(
outcome = rnorm(100),
predictor1 = rnorm(100),
predictor2 = rnorm(100),
predictor3 = rnorm(100)
)

# Fit full linear regression model

full_model <- lm(outcome ~ ., data = data)

# Perform forward selection using stepAIC

library(MASS)
forward_model <- stepAIC(full_model, direction = "forward")

# Summary of the selected model

summary(forward_model)

JSSCACS 2023-24 ROOPA

 20
STATISTICAL COMPUTING AND R PROGRAMMING

OUTPUT:
Call:
lm(formula = outcome ~ predictor1 + predictor2 + predictor3,
data = data)

Residuals:
Min 1Q Median 3Q Max
-2.35541 -0.58837 -0.08408 0.55592 2.31302

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.09952 0.09309 1.069 0.288
predictor1 -0.04102 0.09547 -0.430 0.668
predictor2 -0.12493 0.09719 -1.285 0.202
predictor3 -0.04219 0.08892 -0.474 0.636

Residual standard error: 0.9172 on 96 degrees of freedom

Multiple R-squared: 0.02106, Adjusted R-squared: -0.00953
F-statistic: 0.6885 on 3 and 96 DF, p-value: 0.5613

Creates a sample dataset with an outcome variable (outcome ) and three predictor vari
ables (predictor1, predictor2, predictor3).
 Fits a full linear regression model (full_model) using all predictors.
 Utilizes the stepAIC function from the MASS package to perform forward selection
based on the Akaike Information Criterion (AIC). The direction = "forward" argum
ent specifies forward selection.
 Displays the summary of the selected model resulting from the forward selection proc
ess.
Backward Elimination:
Backward selection is a model selection technique in linear regression where predictors are iteratively
removed from the model based on statistical criteria until no further improvement is observed.
Let's consider a dataset named data with variables X1, X2, X3, and the response variable Y.
Steps:
Data Preparation:
Assume you have a dataset named data containing variables X1, X2,X3, and the response
variable Y.
Backward Selection Implementation:
# Sample data
set.seed(123)
data <- data.frame(
outcome = rnorm(100),
predictor1 = rnorm(100),
predictor2 = rnorm(100),
predictor3 = rnorm(100)
)

JSSCACS 2023-24 ROOPA

 21
STATISTICAL COMPUTING AND R PROGRAMMING

# Fit full linear regression model

full_model <- lm(outcome ~ ., data = data)

# Perform forward selection using stepAIC

library(MASS)
backward_model <- stepAIC(full_model, direction = "backward")
summary(backward_model )
OUTPUT:
Call:
lm(formula = outcome ~ 1, data = data)

Residuals:
Min 1Q Median 3Q Max
-2.39957 -0.58426 -0.02865 0.60141 2.09693

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.09041 0.09128 0.99 0.324

Residual standard error: 0.9128 on 99 degrees of freedom

3. Stepwise selection
Stepwise selection is a technique in linear regression that combines forward and backward
selection methods to iteratively add or remove predictors based on statistical criteria.
 One of the most famous criterion measures is known as Akaike’sInforma tio n
Criterion(AIC). To have noticed this value as one of the columns in the output of
add1 and drop1.For a given linear model, AIC is calculated as follows:

AIC = −2 × L + 2 × (p + 2)
Here, L is a measure of goodness-of- fit named the log-likelihood, and p is the number
of regression parameters in the model, excluding the overall intercept
Source code:
# Sample data
set.seed(123)
data <- data.frame(
outcome = rnorm(100),
predictor1 = rnorm(100),
predictor2 = rnorm(100),
predictor3 = rnorm(100)

JSSCACS 2023-24 ROOPA

 22
STATISTICAL COMPUTING AND R PROGRAMMING

# Fit full linear regression model

stepwise_model <- lm(outcome ~ ., data = data)

# Perform forward selection using stepAIC

library(MASS)
backward_model <- stepAIC(full_model, direction = "both")
summary(backward_model )
OUTPUT:
Call:
lm(formula = outcome ~ 1, data = data)

Residuals:
Min 1Q Median 3Q Max
-2.39957 -0.58426 -0.02865 0.60141 2.09693

Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.09041 0.09128 0.99 0.324

Residual standard error: 0.9128 on 99 degrees of freedom

Linear Residual Diagnostics

In linear model selection, diagnostics play a crucial role in assessing the model's performance
and identifying potential issues or violations of model assumptions.

Some common diagnostics include:

1. Residual Analysis: Examining residuals (difference between observed and predicted

values) to ensure they follow a normal distribution, have constant variance, and
show no clear patterns in their plot against predicted values.
2. Influence and Outlier Detection: Identifying influential data points or outliers that
might significantly affect the model fit. Tools like Cook's distance, leverage, and
studentized residuals help in detecting these.
3. Multicollinearity: Assessing the correlation between predictors to identify
multicollinearity, which can affect the stability and interpretation of coefficients.
4. Homoscedasticity: Checking for constant variance in residuals across different levels
of predictors. A plot of residuals against fitted values can help diagnose this.
5. Normality of Residuals: Verifying whether residuals approximately follow a normal
distribution through techniques like QQ-plots or statistical tests.

JSSCACS 2023-24 ROOPA

 23
STATISTICAL COMPUTING AND R PROGRAMMING

ADVANCED GRAPHICS:

1. Many users are first drawn to R because of its impressive graphical flexibility and
the ease with which you can control and tailor the resulting visuals.
2. Advanced plot customization in statistics involves fine-tuning visual elements of
graphs and charts to convey data effectively.
3. It includes adjusting colors, adding annotations, changing axis scales, incorporating
multiple data series, and utilizing various plot types to enhance the representation of
statistical information.
4. The advanced graph in statistics visual representation of the data.
5. Advanced graph customization allows you to tailor every aspect of your visualizatio ns.
6. This can involve adjusting axis limits, changing tick marks, customizing grid lines,
modifying line styles and marker shapes, incorporating annotations, adjusting legends,
applying color palettes, and even creating interactive or 3D plots.
Characterstics of advanced graphs
1) Color and Style:
Color Palettes: Choose meaningful color schemes for different data elements. Use
tools like color gradients or categorical color palettes.
Line Styles and Markers: Customize the appearance of lines, such as dashed or
dotted lines, and markers for data points.
2) Annotations:
Text Annotations: Add labels, titles, or captions to highlight important features or
provide additional context.
Arrows and Lines: Use arrows or lines with annotations to draw attention to
specific data points or trends.
3) Axis Customization:
Tick Marks and Labels: Adjust the appearance of tick marks and labels on both
the x and y axes.
Logarithmic Scales: Utilize logarithmic scales for axes if data spans multip le
orders of magnitude.
4) Legend:
Positioning: Move the legend to a suitable position (top, bottom, left, right) for
better readability.

JSSCACS 2023-24 ROOPA

 24
STATISTICAL COMPUTING AND R PROGRAMMING

Custom Labels: Provide clear and concise labels for each series in the legend.
5) Faceting and Multiple Plots:
Facet Grids: Use facets to create multiple plots based on different categories,
making it easier to compare subsets of the data.
Arrangement: Adjust the arrangement of subplots to create a coherent and
informative layout.
6) Statistical Summaries:
Error Bars: Include error bars to show variability or uncertainty in data
Regression Lines: Add regression lines or other statistical summaries to illustrate
trends.
7) Background and Grid Lines:
Background Color: Customize the background color to improve contrast and
aesthetics.
Grid Lines: Adjust the appearance of grid lines to guide the viewer's eye.
8) Export and Save Options:
High-Resolution Output: Ensure that exported graphs are of high resolution for
publications or presentations.
Save in Multiple Formats: Save plots in different formats (PDF, PNG, SVG) for
versatility.

Handling the Graphics Device

In advanced graphics using or Manually Opening a New Device to using three built in function
1. dev. new () --------To open the new device plot

2. dev. off () ---------To close the device plot after completion of the task.

3. dev. set () --------- Switching Between Devices to change something in Device 2 without
closing Device 3, use dev. Set

Plot customization: Advanced plot customization in advanced graphs involves fine-tuning visual
elements to convey complex data effectively.

Customizing plots in R can be done using various functions and parameters to modify the appearance
of different plot elements. Here's a basic overview of how you can customize plots in R.

Characteristics of Plot customization:

JSSCACS 2023-24 ROOPA

 25
STATISTICAL COMPUTING AND R PROGRAMMING

1. Titles and Labels

2. Colors and Symbols

3. Adding a Legend

4. Axis Customization
5. Lines

6. Annotations

7. Plot Range
8. Defining a Particular Layout
1.Titles and Labels:
 main: Main title of the plot.
 xlab and ylab: Labels for the x and y-axes.
Example:
plot(x, y, main="Scatter Plot", xlab="X-axis label", ylab="Y-axis label")

output:

2.Colors and Symbols:

col: Color of the points or lines.
pch: Symbol to be used for plotting points.

JSSCACS 2023-24 ROOPA

 26
STATISTICAL COMPUTING AND R PROGRAMMING

Example:
plot(x, y, col="blue", pch=16)
Output:

3.Adding a Legend:
legend: Adding a legend to the plot.
Example:
legend("topright", legend=c("Group 1", "Group 2"), col=c("red", "blue"), pch=16)

output:

4.Axis Customization:
axis: Customize the axis ticks and labels.

JSSCACS 2023-24 ROOPA

 27
STATISTICAL COMPUTING AND R PROGRAMMING

Example:
plot(x, y, xaxt="n") # Turn off x-axis
axis(1, at=c(1, 2, 3), labels=c("Label 1", "Label 2", "Label 3"))
output:

5.Lines:
lines: Add lines to the plot.
abline: Add a straight line.
Example:
lines(x, y, col="green") abline(h=0, v=0,
col="red", lty=2)
output:

6.Annotations:
text: Add text annotations to the plot.
Example:

JSSCACS 2023-24 ROOPA

 28
STATISTICAL COMPUTING AND R PROGRAMMING

text(x, y, labels="Annotation", pos=1, col="purple")

output:

7.Plot Range:

xlim and ylim: Set the range of the x and y-axes.

Example:
plot(x, y, xlim=c(0, 10), ylim=c(-5, 5))
output:

8.Defining a Particular Layout:

The arrangements of plots in a single device using the layout function, which offers more
ways to individualize the panels into which the plots will be drawn.

Example: lay. Mat <- matrix(c(1,3,2,3),2,2)

lay. Mat
[,1] [,2]
[1,] 1 2
[2,] 3 3
layout(mat=lay. Mat)

JSSCACS 2023-24 ROOPA

 29
STATISTICAL COMPUTING AND R PROGRAMMING

layout. show(n=max(lay. Mat))

Output:

Plotting Regions and Margins

Definition : Regions are areas of a Minitab graph that can be resized, relocated, and formatted
with color, fill patterns, edge line types, and more.
 The figure region is the area that contains the space for your axes, their labels, and
any titles. These spaces are also referred to as the figure margins.
 The outer region, also referred to as the outer margins, is additional space around
the figure region that is not included by default but can be specified if it’s needed.
 The margins of a plot can be customized with the mar graphical parameter. These
graphical parameters are vectors of the form c(bottom, left, top, right), where
each element represent the margin of each side of the plot (in margin lines or in
inches). Note that you can access the current values with par("mar").
For any single plot created using base R graphics, there are three regio n s
that make up the image.
1. plot region: The plot region is all you’ve dealt with so far. This is
where your actual plot appears and where you’ll usually be drawing
your points, lines, text, and so on. The plot region uses the user

JSSCACS 2023-24 ROOPA

 30
STATISTICAL COMPUTING AND R PROGRAMMING

coordinate system, which reflects the value and scale of the horizo nta l
and vertical axes.
2. Figure region: The figure region is the area that contains the space
for your axes, their labels, and any titles. These spaces are also referred
to as the figure margins.
3. outer region :The outer region, also referred to as the outer margins,
is additional space around the figure region that is not included by
default but can be spec- ified if it’s needed

1.Plotting Regions: Use rect to draw rectangles or regions on a plot.

Example:
# Draw a rectangle from (1, 2) to (4, 6)

plot(1:10, type="n") # Create an empty

plotrect(1, 2, 4, 6, col="lightblue", border="blue")

output:

2.Plotting Margins:: Margin is the space between the chart border and the
canvas border. You can set the chart margins on any one of the chart's four sides.
Adjust the margins using par(mar=...). The mar parameter specifies the margin sizes in inches
(bottom, left, top, right).
Example:
par(mar=c(5, 4, 4, 2)) # Set margins (bottom, left, top, right)

plot(1:10)
output:

JSSCACS 2023-24 ROOPA

 31
STATISTICAL COMPUTING AND R PROGRAMMING

Special built in function using plot regions and margins.

1.mar (margin lines)

By default par(mar = c(5, 4, 4, 2) + 0.1).

Note that the 0.1 is just for adding an extra space to fit everything inside the plot area.

2.Outer margin(oma )

 The margin in margin lines with oma. The outer margins are specially useful for
adding text to a combination of plots.
 To access the current outer margins with par("oma").
In this example we set

par(oma = c(2, 1, 2, 3) + 0.1),

so two lines are displayed below the plot, one on the left, two on the top and three on the
right.
Output:

JSSCACS 2023-24 ROOPA

 32
STATISTICAL COMPUTING AND R PROGRAMMING

Example:

par(oma=c(1,4,3,2),mar=
4:7) R> plot(1:10)
R>
box("figure",lty=2)
R>
box("outer",lty=3)
Output:

Note:The graphical parameters oma (outer margin) and mar (figure margin)
are used to control these regions and magins like mfrow, they are initialized

JSSCACS 2023-24 ROOPA

 33
STATISTICAL COMPUTING AND R PROGRAMMING

through a call to par before you begin to draw any new plot.
3.Default spacing:default figure margin settings with a call to par in R
Example:

1.par()$oma

[1] 0 0 0 0

2. par()$mar
[1] 5.1 4.1 4.1 2.1

here that oma=c(0, 0, 0, 0)—there is no outer margin set by default. The default
figure margin space is mar=c(5.1, 4.1, 4.1, 2.1)—in other words, 5.1 lines of text
on the bottom, 4.1 on the left and top, and 2.1 on the right.

3.consider the image on the left of Figure 23-4, created in a fresh graphics
device with the following:
plot(1:10)
box(which="figure",lty=2)

4.Custom Spacing: Let’s produce the same plot but with tailored outer margins so that
the bottom, left, top, and right areas are one, four, three, and two lines, respec tively,
and the figure margins are four, five, six, and seven lines.
Example:
par(oma=c(1,4,3,2),mar=4:7)
plot(1:10)

JSSCACS 2023-24 ROOPA

 34
STATISTICAL COMPUTING AND R PROGRAMMING

box("figure",lty=2)
box("outer",lty=3)

mtext("Figure region margins\nmar[ . ]",line=2)

mtext("Outer region margins\noma[ . ]",line=0.5,outer=TRUE)

Output:

provide the text you want written in a character string as the fir s t
argument, and the argument line instructs how many lines of space awa y
from the inside border the text should appear.
5.Clipping:
Controlling clipping allows you to draw in or add elements to the mar- gin regions with
reference to the user coordinates of the plot
Example:

# create margin around plot

par(mar = c(3, 3, 3, 8), xpd = TRUE)

# Draw scatter plot

plot(sample_data$x, sample_data$y, col = sample_data$group)

# Draw legend

JSSCACS 2023-24 ROOPA

 35
STATISTICAL COMPUTING AND R PROGRAMMING

legend("topright", inset = c(-0.3, 0.1), legend = c("Group 1","Group 2"), pch = c(11,12), col
= 1:2)
Output:

Point-and-Click Coordinate Interaction :

Point-and-click coordinate interaction in a graph refers to the capability for users to interact
with plotted data points by clicking or hovering over them to trigger specific actions or to
display additional information related to those points.
 To use the locator() function for point-and-click coordinate interaction. This function
allows you to click on a plot, and it returns the coordinates of the point where you clicked


Features of Point-and-Click Coordinate Interaction :

1. Retrieving Coordinates Silently
2. Visualizing Selected Coordinates
3.Ad Hoc Annotation

1. Retrieving Coordinates Silently:

 The locator command allows you to find and return user coordinates.
 To see how it works, first execute a call to plot(1,1) to bring up a simple plot with a
single point in the middle.
 To use locator, you simply execute the function (with no arguments for defa ult
behavior), which will “hang” the console, without returning you to the prompt.

JSSCACS 2023-24 ROOPA

 36
STATISTICAL COMPUTING AND R PROGRAMMING

Example:
plot(1,1)
locator()
$x
[1] 0.8275456 1.1737525 1.1440526 0.8201909
$y
[1] 1.1581795 1.1534442 0.9003221 0.8630254
2.Visualizing Selected Coordinates
To use locator to plot the points you select as either individual points or as lines
Example:
plot(1,1)
list <- locator(type="o",pch=4,lty=2,lwd=3,col="red",xpd=TRUE) R>

list
$x
[1] 0.5013189 0.6267149 0.7384407 0.7172250 1.0386740 1.2765699
[7] 1.4711542 1.2352573 1.2220592 0.8583484 1.0483300 1.0091491

$y
[1] 0.6966016 0.9941945 0.9636752 1.2819852 1.2766579 1.4891270
[7] 1.2439071 0.9630832 0.7625887 0.7541716 0.6394519 0.9618461

2. Ad Hoc Annotation
The locator function also allows you to place ad hoc annotations, such as lege nd s,
on your plot—remember, since locator returns valid R user coordinates, these
results can directly form the positional argument of most standard annota tio n
functions.
 Example:
plot(survey$Height~survey$Wr.Hnd,pch=16,col=c("gray","black")[as.num eric(survey$
Sex)], xlab="Writing handspan",ylab="Height")
 legend(locator(n=1),legend=levels(survey$Sex),pch=16,col=c("gray","black"))

OUTPUT:

JSSCACS 2023-24 ROOPA

 37
STATISTICAL COMPUTING AND R PROGRAMMING

CUstomizing Traditional R Plots:

 customize traditional plots extensively for instance, in a scatter plot, you can adjust the
colors, sizes, shapes of points, add labels, change axes' appearance, titles, legends, and
much more using various functions within the ggplot2 package or even base R graphics
functions.
 Customizing traditional R plots involves modifying various aspects of the plot to make
it visually appealing and informative. To customize different elements of a plot in R.

Types of CUstomizing Traditional R Plots

1. Graphical Parameters for Style and Suppression.
2. Customizing Boxes
3. Customizing Axes
1) Graphical Parameters for Style and Suppression.
Graphical parameters in R, managed by the par() function, allow you to control various aspects
of plot appearance and layout. They enable customization of plots by modifying parameters
such as margins, axis appearance, text size, colors, and more.

Example:

hp <- mtcars$hp

mpg <- mtcars$mpg

wtcex <- mtcars$wt/mean(mtcars$wt)

plot(hp,mpg,cex=wtcex)

plot(hp,mpg,cex=wtcex,xaxs="i",yaxs="i")

Output:

JSSCACS 2023-24 ROOPA

 38
STATISTICAL COMPUTING AND R PROGRAMMING

:
 let’s plot MPG against horsepower (from the ready-to-use mtcars data set)
and set each plotted point to be sized propor tionally to the weight of each
car.
 This plot is almost the same as the default, but note now that there’s no padding
space at the end of the axes; the most extreme data points sit right on the axes.
2.Customizing Boxes:
To add a box specific to the current plot region in the active graphics device, you use box
and specify its type with bty.

box(bty="u")
Example:

box(bty="]",lty=2,col="gray")

Output:

Various box configurations added to the mtcars scatterplot

JSSCACS 2023-24 ROOPA

 39
STATISTICAL COMPUTING AND R PROGRAMMING

 The bty argument is supplied a single character: "o" (default), "l", "7", "c", "u",
"]", or "n". The help file entry for bty

 The resulting box boundaries will follow the appearance of the corresponding
uppercase letter, with the exception of "n"

3.Customizing Axes
The axis function allows you to control the addition and appearance of an axis on any of the
four sides of the plot region.
The first argument it takes is side, provided with a single integer: 1 (bottom), 2 (left), 3
(top), or 4 (right).
Syntax:

axis(side, at=NULL, labels=TRUE)

Parameters:

side: It defines the side of the plot the axis is to be drawn on possible values such as
below, left, above, and right.
at: Point to draw tick marks
Example:

x <- 1:5; y = x * x
plot(x, y, axes =
FALSE)
# Calling the axis() function
axis(side = 1, at = 1:5, labels =
LETTERS[1:5]) axis(3)

Output: Output:

JSSCACS 2023-24 ROOPA

 40
STATISTICAL COMPUTING AND R PROGRAMMING

Specialized Text and Label Notation :

Immediately accessible tools for controlling fonts and displaying special
notation, such as Greek symbols and mathematical expressions.
Features of Specialized Text and Label Notation :

1.Font:
The displayed font is controlled by two graphical parameters. family for the
specific font family and font, an integer selector for controlling bold and italic
typeface.
 Available fonts depend on both your operating system and the graphics
 Three generic families— “sans” the default, “serif”, and “mono”— are
always available
 These are paired with the four possible values of font—1 (normal text, default), 2
(bold), 3 (italic), and 4 (bold and italic).

Example:
par(mar=c(3,3,3,3))
plot(1,1,type="n",xlim=c(-1,1),ylim=c(0,7),xaxt="n",yaxt="n",ann=FALSE)
text(0,6,label="sans text (default)\nfamily=\"sans\", font=1")
text(0,5,label="serif text\nfamily=\"serif\", font=1", family="serif",font=1)
text(0,4,label="mono text\nfamily=\"mono\", font=1", family="mono",font=1)
text(0,3,label="mono text (bold, italic)\nfamily=\"mono\", font=4", family="mono",font=4)
OUTPUT:

JSSCACS 2023-24 ROOPA

 41
STATISTICAL COMPUTING AND R PROGRAMMING

Displaying font styles through use of the family and font graphical parameters
2. Greek Symbols
For statistically or mathematically technical plots, annotation may occasionally require
Greek symbols or mathematical markup.

 display these using the expression function.

Example:
par(mar=c(3,3,3,3))
plot(1,1,type="n",xlim =c(-1,1),ylim =c(0.5,4.5),xaxt="n",yaxt="n", ann=FALSE)
text(0,4,label=expression(alpha),cex=1.5)
text(0,3,label=expression(paste("sigma: ",sigma,"Sigma: ",Sigma)),
family="mono",cex=1.5)
text(0,2,label=expression(paste(beta," ",gamma," ",Phi)),cex=1.5)
text(0,1,label=expression(paste(Gamma,"(",tau,") = 24 when ",tau," = 5")),
family="serif",cex=1.5)
title(main=expression(paste("Gr",epsilon,epsilon,"k")),cex.main=2)
Output:

JSSCACS 2023-24 ROOPA

 42
STATISTICAL COMPUTING AND R PROGRAMMING

3.Mathematical Expressions:
Formatting entire mathematical expressions to appear in R plots is a bit mor e
complicated and is reminiscent of using markup languages

To create the image, I first defined four expression objects as follows:

Example:
expr1 <- expression(c^2==a[1]^2+b[1]^2)
expr2 <- expression(paste(pi^{x[i]},(1-pi)^(n-x[i])))
expr3 <- expression(paste("Sample mean:",italic(n)^{-
1},sum(italic(x)[italic(i)], italic(i)==1, italic(n))
==frac(italic(x)[1]+...+italic(x)[italic(n)], italic(n))))
expr4 <- expression(paste("f(x","|",alpha,",",beta,")"==frac(x^{alpha-1}~(1-x)^{beta-1},
B(alpha,beta))))
Output:

JSSCACS 2023-24 ROOPA

 43
STATISTICAL COMPUTING AND R PROGRAMMING

Examples of typesetting mathem atical expressions in R plots

Defining colors and plotting in higher dimensions:
Colors:The colors using various methods, such as specifying named colors, hexadecimal
color codes, RGB values, HSL values, among others. The choice of color representation
depends on your preference and the context of your visualization.
1.Named Colors:
R provides a range of named colors you can use directly. Some common ones include:
 Red: "red"
 Blue: "blue"
 Green: "green"
 Yellow: "yellow"
 Black: "black"
 White: "white"
2.Hexadecimal Color Codes:
Hexadecimal color codes are a common way to define colors. They consist of a # symbol
followed by a combination of six characters (0-9 and A-F) representing the intensity of red,
green, and blue (RGB) channels
For example:
1. Red: "#FF0000"

JSSCACS 2023-24 ROOPA

 44
STATISTICAL COMPUTING AND R PROGRAMMING

2. Green: "#00FF00"
3. Blue: "#0000FF"
4. Orange: "#FFA500"
5. Purple: "#800080"
3.RGB Values:
You can define colors using RGB values, specifying the intensity of red, green, and blue channels
individually within the range of 0 to 1.
1. Red: rgb(1, 0, 0)
2. Green: rgb(0, 1, 0)
3. Blue: rgb(0, 0, 1)
4. Yellow: rgb(1, 1, 0)
5. Purple: rgb(0.5, 0, 0.5

4.HSL Values
HSL (Hue, Saturation, Lightness) values can also be used to define colors. The hue value represents
the color (0-360), saturation represents intensity (0-1), and lightness represents the brightness (0-1).
For example:
1. Red: hsl(0, 1, 0.5)
2. Green: hsl(120, 1, 0.5)
3. Blue: hsl(240, 1, 0.5)
4. Yellow: hsl(60, 1, 0.5)
5. Purple: hsl(300, 1, 0.5)
Example:
# Using different color representations in a plot

my_color <- "red"

plot(1:10, col=my_color, pch=16)
plot(1:10, col = "blue", main = "Named Color")

my_rgb_color <- rgb(0.2, 0.5, 0.8)

plot(1:10, col=my_rgb_color, pch=16)
plot(1:10, col = "#FFA500", main = "Hexadecimal Color")

my_hex_color <- "#FFA500" # Orange

plot(1:10, col=my_hex_color, pch=16)
plot(1:10, col = rgb(0, 1, 0), main = "RGB Color")

plot(1:10, col = hsl(240, 1, 0.5), main = "HSL Color")

JSSCACS 2023-24 ROOPA

 45
STATISTICAL COMPUTING AND R PROGRAMMING

OUTPUT:

Plotting in higher dimensions:

Plotting in higher dimensions often involves techniques like 3D plots, color mapping, facet
grids, or even interactive visualizations to represent more than three dimensions effectively

JSSCACS 2023-24 ROOPA

 46
STATISTICAL COMPUTING AND R PROGRAMMING

Characteristics of Plotting in higher dimensions:

1.Red-Green-Blue Hexadecimal Color Codes:
 To specifying colors in plots, your instruction to R so far has been given either in
the form of an integer value from 1 to 8 or as a character
 One of the most common methods of color specification is to specify differ e nt
saturations or intensities of three primaries—red, green, and blue (RGB)
Example:
1.palette()
[1] "black" "red" "green3" "blue" "cyan" "magenta"
[7] "yellow" "gray"
The col argument lets you select one of eight colors when you supply it an integer from
1 to 8
2. rgb(t(col2rgb(c("black","green3","pink"))),maxColorValue=255)
[1] "#000000" "#00CD00" "#FFC0CB"

2.2D Plots: ‘2D’ stands for 2-dimensional and a 2D line is a line that is moved in 2-
dimensions. A line in 2D means that we could move in forward and backward direction but
also in any direction like left, right, up, down.
Scatter Plot:

Using basic plot()

function. x <-
rnorm(100)
y <- rnorm(100)
plot(x, y, col="blue", pch=16)
output:

3.Surface Plot: Surface plots are diagrams of three-dimensional data. Rather than showing the
individual data points, surface plots show a functional relationship between a designated

JSSCACS 2023-24 ROOPA

 47
STATISTICAL COMPUTING AND R PROGRAMMING

dependent variable (Y), and two independent variables (X and Z). The plot is a companion plot
to the contour plot.

Example:
x <- seq(-10, 10,
length=100) y <- seq(-10,
10, length=100)
z <- outer(x, y, function(x, y) sin(sqrt(x^2 + y^2)))
persp(x, y, z, theta=30, phi=20, col="lightblue", shade=0.5)
output:

REPRESENTING AND USING COLOR:

 One of the most common methods of colour specification is to specify differe nt
saturations or intensities of three primaries—red, green, and blue (RGB)—which are
then mixed to form the resulting target colour.
 To express these values in (R, G, B) order; the result is commonly referred to as a
triplet. For example, (0;0;0) represents pure black, (255;255;255) represents pure
white, and (0;255;0) is full green.
1.Palette():A palette in Rit is simply a built in finction holds the few colours.
palette()
[1] "black" "red" "green3" "blue" "cyan" "magenta" "yellow"
[8] "gray
2.Defining your own palettes
If you want to make your own palette, you can just create your own vector of colours.
Example:
colors <- c("#A7A7A7", "dodgerblue", "firebrick", "forestgreen", "gold")

hist(discoveries, col = colors)

JSSCACS 2023-24 ROOPA

 48
STATISTICAL COMPUTING AND R PROGRAMMING

OUPUT:

2. Built-in Palettes
 Being able to implement your own RGB colors is most useful when you need
many colors, the collection of which is referred to as a palette.
 There are a number of color palettes built into the base R installa- tion. These
are defined by the functions rainbow, heat.colors, terrain.colors, topo.colors,
cm.colors, gray.colors, and gray.
Example:
N <- 600
rbow <- rainbow(N)
heat <- heat.colors(N)
terr <- terrain.colors(N)
topo <- topo.colors(N)
cm <- cm.colors(N)
gry1 <- gray.colors(N)
gry2 <- gray(level=seq(0,1,length=N))
dev.new(width=8,height=3)
par(mar=c(1,8,1,1))
plot(1,1,xlim=c(1,N),ylim=c(0.5,7.5),type="n",xaxt="n",yaxt="n",ann=FALSE)
points(rep(1:N,7),rep(7:1,each=N),pch=19,cex=3,col=c(rbow,heat,terr,topo,cm,gry1,gry2))
OUTPUT:

JSSCACS 2023-24 ROOPA

 49
STATISTICAL COMPUTING AND R PROGRAMMING

Showcasing the color ranges of the built-in palettes, with default limits used in gray.colors.
3.Color Palettes:
R provides color palettes that allow you to use a sequence of colors.
Example:
# Example using a color palette
colors <- rainbow(10) # Generates 10 colors of the

rainbow barplot(1:10, col=colors)

output:

4. Color Functions:
R has functions for creating color gradients or interpolating colors.
Example:

library(RColorBrewer)
colors <- colorRampPalette(brewer.pal(9,

"Blues"))(10) pie(rep(1, 10), col=colors)

JSSCACS 2023-24 ROOPA

 50
STATISTICAL COMPUTING AND R PROGRAMMING

output:

5.Color Scales:
R supports various color scales, like viridis, magma, etc.
Example:
library(viridis)
plot(1:10, col=viridis(10), pch=16)
output:

3D SCATTER PLOTS.:
creating 3D scatterplots, which allow you to plot raw observations based on thre e
continuous variables at once, as opposed to only two in a conventional 2D scatterplot

 3D scatterplots allow you to plot raw observations based on three continuous

variables at once.
 The syntax of the scatterplot3d function is similar to the default plot function.
 In the latter, you supply a vector of x- and y-axis coordinates; in the former, you
merely supply an additional third vector of values providing the z-axis
coordinates.

Syntax:

scatterplot3d(x, y=NULL, z=NULL)

JSSCACS 2023-24 ROOPA

 51
STATISTICAL COMPUTING AND R PROGRAMMING

x, y, z are the coordinates of points to be plotted.

Example:

scatterplot3d(iris[,1:3], main="3D Scatter Plot", xlab = "Sepal Length(cm)", ylab = "Sepal

Width (cm)", zlab = "Petal Length (cm)")
Output:

Packages Using 3D scatterplot

 Creating 3D scatter plots in R can be achieved using different packages.

 One popular package for this purpose is scatterplot3d. Here's a basic example of
how to create a 3D scatter plot using this package:
Example:
#Install and load the scatterplot3d package
install.packages("scatterplot3d")
library(scatterplot3d)
# Generate some random data set.seed(123)
x <- rnorm(100)
y <- rnorm(100)
z <- rnorm(100)

# Create a 3D scatter plot

scatterplot3d(x, y, z, color="blue", pch=16, main="3D Scatter Plot")

JSSCACS 2023-24 ROOPA

 52
STATISTICAL COMPUTING AND R PROGRAMMING

OUTPUT:

Fifth Module Questions:

I. Two marks Questions:
1.What is simple linear regression.
2.Define plotting regions and margins.
3.What is 3D scatter plots.
4.Define regions and margins.
5.what are all the specialized text
II. Three marks Questions:
1. define simple linear regression with example.
2. What is prediction? Explain prediction /confidence interval.
3. Explain plot customization.
4. Discuss specialized text and label notation.
5. Explain 3D scatter plots
III. Five marks Questions:
1.what is simple linear regression in detail.
2.Explain plot customization
3.Defining colours and plotting in higher dimensions.
4. write a r program to create an any application of linear regression in multivariate
context for predictive purpose.
5.Explain representing and using colour.
IV. Ten marks Questions:
1.Discuss simple linear regression.
2.Briefly explain Multi linear regression.
3.Explain plotting regions and margins.
4.Explain customizing traditional R plots.
5.Explain 3D scatter plot in details.

JSSCACS 2023-24 ROOPA

 53
STATISTICAL COMPUTING AND R PROGRAMMING

JSSCACS 2023-24 ROOPA