i
Search icon

R Programming Complete Tutorial

Self Learning   /   Study Material   /   Heart Risk Analysis using SVM
Previous
Next

Heart Risk Analysis using SVM

Coronary heart diseases remain one of the leading causes of death all over the world. One of the biggest contributors to coronary disease is a lack of commitment to a heart-healthy lifestyle and the consequences associated with it.

This project aims at early detection that means finding out whether the patients have the risk of coronary heart disease in the next ten years. We have used the Data Set, heartdata for this analysis.

Step 1: We are going to split the dataset into training and test set.

Step 2: Model Training using a training set

In this section, we will train our model using svm() function. The syntax is available in the below section. 

svm () Syntax:

We will use the same Training dataset used in the Logistic regression model. Now we will use the below command to train the SVM model.

In svm () function:

In our experiment, we used the svm() function with the kernel as a polynomial. We did our test with different cost functions (.001, .01, 1, and 10) and observed that svm() function with .001 cost was giving us the best result. We have stored that result in the svm_Linear variable.

Detail of svm_Linear:

Step 3: Model Prediction using Test data set.

The caret package provides the predict () method for predicting results. We are passing two arguments. Its first parameter is our trained model, and the second parameter, "newdata” holds our testing data frame. The predict() method returns a list. We are saving it in a predict_svm variable.


Step 4: Model Evaluation using Confusion Matrix and ROC Curve:

Confusion Matrix for SVM

In the case of Support Vector Machine, we will use confusionmatrix function from the caret package. In this function, we are going to use our table function. It will give us the detail of the matrix, including different measurements like sensitivity, specificity, and many more. In our test, we got only 1 true positive and no false-positive values. 

-ROC Curve

AUC - ROC curve is a performance measurement for the classification problem at various threshold settings. ROC is a probability curve, and AUC represents the degree or measure of separability. This shows how much the model is capable of differentiating between classes. Higher the AUC, better the model is at predicting 1s as 1s and 0s as 0s. Through contrast, the higher the AUC, the easier the model is to distinguish between patients at risk for heart failure and no risk.

ROC-AUC Curve for SVM:

© All rights Reserved | SelfLearning.io