The one-class SVM is designed to detect anomalies or outliers in a dataset, so it is trained to recognize only one class. When testing with a dataset that contains both classes, the one-class SVM will classify all instances as either the target class or outliers. Therefore, the confusion matrix will only show the performance of the model in detecting the target class and will not include the other class.
If you want to evaluate the performance of the model on both classes, you would need to use a different classification algorithm that supports multi-class classification, such as a Support Vector Machine (SVM) with a multi-class extension or a different algorithm altogether.
For more information on one-class SVM, please refer to the documentation below:
Hope this helps.
