This report demonstrates how to implement a Support Vector Machine (SVM) model in R using the iris dataset. We’ll go through data loading, preprocessing, model training, evaluation, and tuning.

SVM is a supervised machine learning algorithm used for classification and regression. It works by finding the best hyperplane to separate different classes in the feature space.

We use the e1071 package in R, which includes the svm() function to build and tune SVM models.

Install e1071 library if not already installed

if (!require("e1071")) {
  install.packages("e1071", dependencies = TRUE)
}

Load e1071 library

library(e1071)

Load built-in iris dataset

data(iris)

The iris dataset contains 150 flower observations with 4 numeric features and one categorical target (Species).

Display the first six rows of the dataset

head(iris)

##   Sepal.Length Sepal.Width Petal.Length Petal.Width Species
## 1          5.1         3.5          1.4         0.2  setosa
## 2          4.9         3.0          1.4         0.2  setosa
## 3          4.7         3.2          1.3         0.2  setosa
## 4          4.6         3.1          1.5         0.2  setosa
## 5          5.0         3.6          1.4         0.2  setosa
## 6          5.4         3.9          1.7         0.4  setosa

Display the statistical summary of the dataset

summary(iris)

##   Sepal.Length    Sepal.Width     Petal.Length    Petal.Width   
##  Min.   :4.300   Min.   :2.000   Min.   :1.000   Min.   :0.100  
##  1st Qu.:5.100   1st Qu.:2.800   1st Qu.:1.600   1st Qu.:0.300  
##  Median :5.800   Median :3.000   Median :4.350   Median :1.300  
##  Mean   :5.843   Mean   :3.057   Mean   :3.758   Mean   :1.199  
##  3rd Qu.:6.400   3rd Qu.:3.300   3rd Qu.:5.100   3rd Qu.:1.800  
##  Max.   :7.900   Max.   :4.400   Max.   :6.900   Max.   :2.500  
##        Species  
##  setosa    :50  
##  versicolor:50  
##  virginica :50  
##                 
##                 
## 

Split the data: 70% for training, 30% for testing. This allows evaluation of model performance on unseen data.

Set seed for reproducibility

set.seed(123)

Split 70% for training and 30% for testing

index <- sample(1:nrow(iris), 0.7 * nrow(iris))
train_data <- iris[index, ]
test_data <- iris[-index, ]

Display the training dataset dimensions

dim(train_data)

## [1] 105   5

Display the testing dataset dimensions

dim(test_data)

## [1] 45  5

We train an SVM model using a linear kernel. The Species ~ . formula means we use all other variables to predict the species.

model <- svm(Species ~ ., data = train_data, kernel = "linear")

Display the statistical summary of the SVM model

summary(model)

## 
## Call:
## svm(formula = Species ~ ., data = train_data, kernel = "linear")
## 
## 
## Parameters:
##    SVM-Type:  C-classification 
##  SVM-Kernel:  linear 
##        cost:  1 
## 
## Number of Support Vectors:  24
## 
##  ( 2 10 12 )
## 
## 
## Number of Classes:  3 
## 
## Levels: 
##  setosa versicolor virginica

Use the trained model to predict flower species in the test dataset.

Make predictions

predictions <- predict(model, test_data)

Display the first six predictions

head(predictions)

##      1      2      3      5     11     18 
## setosa setosa setosa setosa setosa setosa 
## Levels: setosa versicolor virginica

Confusion matrix helps us measure the model’s performance. We also calculate accuracy to evaluate correctness.

conf_mat <- table(Predicted = predictions, Actual = test_data$Species)
conf_mat

##             Actual
## Predicted    setosa versicolor virginica
##   setosa         14          0         0
##   versicolor      0         17         0
##   virginica       0          1        13

Display the accuracy

accuracy <- sum(diag(conf_mat)) / sum(conf_mat)
paste("Accuracy:", round(accuracy * 100, 2), "%")

## [1] "Accuracy: 97.78 %"

Use cross-validation to find the best cost and gamma parameters for improved performance with a radial kernel.

tuned_model <- tune(svm, Species ~ ., data = train_data,
                    kernel = "radial",
                    ranges = list(cost = c(0.1, 1, 10),
                                  gamma = c(0.5, 1, 2)))

Display the statistical summary of the tuned model

summary(tuned_model)

## 
## Parameter tuning of 'svm':
## 
## - sampling method: 10-fold cross validation 
## 
## - best parameters:
##  cost gamma
##     1   0.5
## 
## - best performance: 0.04727273 
## 
## - Detailed performance results:
##   cost gamma      error dispersion
## 1  0.1   0.5 0.11363636 0.08487554
## 2  1.0   0.5 0.04727273 0.04994028
## 3 10.0   0.5 0.05727273 0.04943204
## 4  0.1   1.0 0.19181818 0.13001730
## 5  1.0   1.0 0.05636364 0.04863613
## 6 10.0   1.0 0.07545455 0.05861111
## 7  0.1   2.0 0.33545455 0.14928357
## 8  1.0   2.0 0.06545455 0.04531291
## 9 10.0   2.0 0.07545455 0.03998393

Evaluate performance again after tuning. The new model should be more accurate or generalizable.

Select the best model from the tuned model

best_model <- tuned_model$best.model

Make predictions

tuned_predictions <- predict(best_model, test_data)

Create Confusion Matrix table

tuned_conf_matrix <- table(Predicted = tuned_predictions, 
                           Actual = test_data$Species)

Calculate and display the Accuracy

tuned_accuracy <- sum(diag(tuned_conf_matrix)) / 
  sum(tuned_conf_matrix)
paste("Tuned Accuracy:", round(tuned_accuracy * 100, 2), "%")

## [1] "Tuned Accuracy: 97.78 %"

Accuracy Comparison

accuracy_data <- data.frame(
  Model = c("Original", "Tuned"),
  Accuracy = c(round(accuracy, 4), round(tuned_accuracy, 4))
)

# Display the accuracy comparison in a table
kable(accuracy_data)

Tuning these parameters can greatly improve model performance.

• kernel – function that transforms data (e.g., “linear”, “radial”)

• cost – controls margin width and misclassification penalty

• gamma – defines influence of a data point (used in non-linear kernels)

This workflow applies to both small datasets like iris and real-world datasets, such as those used in healthcare analytics.

• Loaded data and trained an SVM model

• Evaluated accuracy and confusion matrix

• Tuned hyperparameters for improved performance