Virtual Labs

Predicting Credit Card Fraud using Support Vector Machine

What is the primary goal of SVM when applied to credit card fraud detection?

a: To encrypt transaction data Explanation

Explanation

b: To classify transactions as fraudulent or legitimate Explanation

Explanation

c: To speed up payment processing Explanation

Explanation

d: To generate new transaction records Explanation

Explanation

In SVM terminology, the data points that define the position of the maximum-margin hyperplane are called:

a: Outliers Explanation

Explanation

b: Slack vectors Explanation

Explanation

c: Support vectors Explanation

Explanation

d: Boundary points Explanation

Explanation

Which kernel is most frequently recommended for real-world credit card fraud detection with SVM?

a: Linear Explanation

Explanation

b: Polynomial Explanation

Explanation

c: Radial Basis Function (RBF) Explanation

Explanation

d: Sigmoid Explanation

Explanation

Why is accuracy a misleading metric for evaluating fraud detection models?

a: Because fraud rates are extremely low (severe class imbalance) Explanation

Explanation

b: Because SVM cannot calculate accuracy Explanation

Explanation

c: Because accuracy requires balanced classes to be meaningful Explanation

Explanation

d: Because fraud data has no ground truth Explanation

Explanation

What is the main role of the regularization parameter C in soft-margin SVM?

a: To control the shape of the RBF kernel Explanation

Explanation

b: To balance margin maximization versus training error minimization Explanation

Explanation

c: To scale the input features Explanation

Explanation

d: To determine the number of support vectors directly Explanation

Explanation

In highly imbalanced fraud detection problems, which evaluation metric is generally preferred over ROC-AUC?

a: Accuracy Explanation

Explanation

b: Precision-Recall AUC (PR-AUC) Explanation

Explanation

c: F1-score of the majority class Explanation

Explanation

d: Mean Absolute Error Explanation

Explanation