GCP Generative AI Leader Practice Question 32: Unlabeled Customer Clustering
Google Cloud Generative AI Leader · Test 4 · question 2 of 10
A marketing team has years of purchase history for millions of customers but no predefined segments or category labels. They want a model to discover natural groupings of customers with similar buying behavior on its own, so the team can design a separate campaign for each group that emerges. Which machine learning approach fits this goal?
Show answer and explanation
Answer:
The key facts are that no labels exist and the team wants the groupings themselves to emerge from the data. Unsupervised learning trains on unlabeled data and finds hidden structure, most commonly through clustering, which groups similar data points, such as customers with similar purchasing behavior, without anyone specifying the categories beforehand.
Option B is wrong because supervised learning needs labeled examples, a known "correct" segment for each customer, and the stem says no such labels exist; inventing them would defeat the goal of letting the groupings emerge. Option C describes reinforcement learning, which trains an agent through actions and rewards over time, not a one-time grouping of historical records. Option D confuses the later use of the segments (writing campaign copy, a generative task) with the task actually being asked for here, which is discovering the segments in the first place.
Reference: What is unsupervised learning?