Machine learning projects focused on discovering hidden structure
through clustering, segmentation, and dimensionality reduction.
Project Focus
Unsupervised learning is useful when the goal is to discover patterns without a predefined target variable. These projects can reveal customer segments, behavioral groups, natural clusters, anomalies, or lower-dimensional structure in complex datasets.
My unsupervised learning projects focus on careful data preparation, feature scaling, cluster validation, visual interpretation, and clear explanations of what the discovered groups mean in practice.
Motivation-Based Participant Segmentation
Unsupervised learning project that explores whether participants can be grouped by similar motivation profiles and whether motivation categories form interpretable response-pattern families.
Methods used: EDA, feature scaling, K-Means, hierarchical clustering, DBSCAN, PCA visualization, silhouette analysis, dendrogram interpretation, and cluster profiling.
Main insight: the participant structure is gradual rather than sharply separated, with one broad main group, a smaller lower drive/exploration profile, and a few especially unusual participants confirmed across methods.
Next Unsupervised Work
This section begins with a complete motivation-based participant segmentation project and will continue growing with additional clustering, anomaly detection, and dimensionality reduction projects.