FRM Part I · FRM Exam Part I · Machine-Learning Methods

A bank's compliance team has millions of unlabelled transaction records and wants an algorithm to group them into segments of similar behaviour, without predefined categories, so that unusual segments can be reviewed. Which approach is most appropriate?

K-means clustering is most appropriate because the data are unlabelled and the goal is to discover groups of similar transactions. The other options are supervised methods that need a known target variable, which this dataset does not provide.

ALogistic regression trained on fraud labels
BK-means clusteringCorrect
CA classification tree trained on known outcomes
DRidge regression of transaction size on account age

Explanation

With no labels and a goal of finding natural groupings, an unsupervised clustering method such as K-means fits. Logistic regression, classification trees and ridge regression all require a labelled target variable, so they are supervised methods.

Did you get it right without looking?

One question tells you little. A timed set on Machine-Learning Methods shows your real accuracy, how long you take and where you lose marks.

Practise Machine-Learning Methods Free diagnostic mock

More Machine-Learning Methods questions