Explore how models react to data. Each simulation is interactive, lightweight, and designed to make assumptions visible—so you can see how decisions change outcomes.
Drag points and compare least‑squares vs. robust fits with instant residual updates.
Launch simulation →Draw repeated samples from skewed, bounded, or bimodal populations and watch the means converge to normal.
Launch simulation →Adjust the soft‑margin strength and watch the boundary shift while support vectors update in real time.
Launch simulation →Place binary points, fit a logistic curve, and explore how cutoff probability changes misclassifications.
Launch simulation →Move a query point through feature space and see the nearest-neighbor vote update in real time.
Launch simulation →Build a tiny network and watch activations and decision regions change as you tweak weights.
Launch simulation →See how sigmoid and ReLU respond to the same input as you sweep through w · x + b.
Launch simulation →Increase polynomial degree and watch training error fall while test/validation error curves back up.
Launch simulation →Cycle through folds, watch the model refit, and track how validation error changes across splits.
Launch simulation →Outer folds estimate generalization while inner folds select polynomial degree.
Launch simulation →Seed centroids, iterate, and watch clusters converge while inertia updates.
Launch simulation →Explore linkage choices and watch dendrogram merges reshape clusters.
Launch simulation →Simulations are grouped by model family: classification, regression, clustering, and probabilistic inference. New items will appear as the course progresses.
Have a model you want to see visualized? Share your idea through Microsoft Forms.
Open suggestion form →