In PMI-CPMAI, selecting the appropriate machine learning approach starts with clarifying the type of question being asked of the data. When upper management wants to “see if there are any patterns and insights that can be discovered from customer data” without predefined labels or outcomes, this maps directly to unsupervised learning.

Unsupervised learning techniques—such as clustering, dimensionality reduction, and association rule mining—are used to uncover hidden structure, segments, or relationships in data where no target variable is specified. PMI-CPMAI training descriptions highlight using such approaches in discovery phases to identify segments, behavioral groupings, or natural patterns that can later inform strategy, product design, or subsequent supervised models.

Reinforcement learning (option C) focuses on agents learning via rewards and penalties through interaction with an environment, which does not fit this “exploratory pattern discovery” objective. Saying “all would work equally well” (option A) contradicts PMI-style guidance, which requires fit-for-purpose selection of AI techniques based on problem framing and data characteristics. Therefore, for discovering patterns and structure in customer data without pre-labeled outcomes, Unsupervised Learning (option B) is the correct choice in line with PMI-CPMAI principles.