Skip to content
Bosaj edited this page Aug 23, 2026 · 1 revision

FAQ

Why 5 clusters and not some other number? The elbow method plot (inertia vs. number of clusters) flattens out around k=5 for the income/spending pairing, and five segments also produce the cleanest business story (see Methodology).

Why cluster on income vs. spending score for the final model, instead of age? Income and spending score together produce visually and commercially distinct groups (e.g. "high income, high spending"), which is more directly actionable for a retail business than age-based clusters.

Is K-means the right algorithm here? For roughly spherical, similarly-sized clusters like these — which is what this dataset produces — K-means is a reasonable and standard choice. It wouldn't be ideal for irregularly shaped or highly imbalanced clusters, but that's not the case here.

What does "Spending Score" actually mean? It's a score from 1–100 assigned by the mall based on customer behavior and purchasing data, as defined by the dataset's creators — not a raw currency amount.

Does the notebook use the gender column in clustering? It's explored during EDA (gender breakdown) but the final clustering models use only the numeric attributes (age, income, spending score).

Does CI verify the cluster assignments or the elbow plot? No — CI only checks notebook structure and dependency installability. It doesn't execute the clustering cells.

Clone this wiki locally