A heatmap represents a one-dimensional or spatial distribution of values, often derived from discrete, sparse measurements that are transformed into a continuous field. This transformation—commonly achieved through methods such as kriging, kernel density estimation (KDE), or other interpolation techniques—creates a smooth surface from scattered data points, revealing gradients, clusters, and anomalies that might otherwise remain hidden.

While these computational methods are often black boxes, visual analytics allows users to open them: to explore parameters, evaluate uncertainty, and understand how algorithmic choices shape the resulting surface. Through interactive visualization, users can iteratively adjust model assumptions, observe their effects in real time, and combine computational inference with domain knowledge.

In this sense, heatmaps are not just visual summaries—they become interfaces for reasoning about data continuity, density, and spatial relationships. By exposing the analytic process itself, visual analytics turns heatmap generation from a static output into a dynamic, knowledge-driven exploration of how discrete data gives rise to continuous insight.

See the video below…