You have a dataset of customer shopping behavior with thousands of features and many numeric columns per customer. You want to quickly find natural groupings among customers and visualize the result. What should you do?

t-SNE (t-Distributed Stochastic Neighbor Embedding) is a dimensionality reduction technique particularly effective for visualizing high-dimensional data in a lower-dimensional space (typically 2D or 3D) while preserving local structures. By embedding the numeric features with t-SNE, you can map the thousands of features to two dimensions, and a scatter plot will then visually represent the natural groupings (clusters) of customers. Running k-means with an elbow plot helps determine the optimal number of clusters but doesn't directly visualize the groupings in a reduced-dimensional space. A line chart is unsuitable for visualizing clusters of points. Box plots for each numeric feature within clusters help understand feature distributions per cluster but don't provide an overall visualization of customer groupings in a reduced space.