A data mining algorithm is unstable if
A) test set accuracy depends on the ordering of test set instances.
B) the algorithm builds models unable to classify outliers.
C) the algorithm is highly sensitive to small changes in the training data.
D) test set accuracy depends on the choice of input attributes.

Question

Quizplus · Accepted Answer

An unstable algorithm is highly sensitive to small changes in the training data, meaning that small variations in the training data can result in very different models being generated. This can lead to unpredictable and inconsistent classification results, making the algorithm unreliable. The other options presented do not necessarily describe an unstable algorithm.

A Data Mining Algorithm Is Unstable If