Visual Data Lab

Explore interactive experiments that make statistical models and machine-learning algorithms easier to understand through visualization.

Explore data, statistical models, and machine-learning algorithms through small interactive experiments.

A simulated ARMA time series with autocorrelation and partial autocorrelation charts.

ARMA Process Simulation
statistics
simulation
time series

See how AR and MA settings change a simulated time series and its autocorrelation patterns.

A decision-tree experiment with a classification boundary, tree diagram, and model diagnostics.

Decision Tree
machine learning
classification
simulation

Build synthetic classification data and watch a decision tree carve the feature space into regions.

High-dimensional data shown in three dimensions and as PCA, Isomap, t-SNE, and UMAP projections.

Dimensionality Reduction
machine learning
visualization
mathematics
runtime-server

Compare PCA, Isomap, t-SNE, and UMAP on simulated and real high-dimensional datasets.

An income distribution, a Lorenz curve with its Gini, a pairwise difference heatmap, and income by percentile.

Income Inequality
statistics
simulation

Explore income distributions, Lorenz curves, and pairwise differences in theoretical and simulated Gini calculations.

Clustered points, centroids, and convergence diagnostics in the K-means experiment.

K-means Step by Step
machine learning
clustering
simulation

Move through each K-means iteration to see centers, assignments, and convergence on generated clusters.

Original and compressed images with sampled color clusters in the K-means experiment.

K-means on Images
machine learning
clustering
visualization
runtime-server

Apply K-means to image colors and compare the original image with a compressed color version.

Simulated classes, a logistic-regression decision boundary, and model diagnostics.

Logistic Regression
machine learning
classification
regression

Change noise, polynomial terms, and class structure to see how logistic regression shapes decision boundaries.

A three-dimensional Lorenz attractor with projection plots and parameter controls.

Lorenz Attractor
mathematics
simulation
visualization

Adjust the Lorenz parameters and watch how small changes produce chaotic 3D trajectories.

Matrix inputs and visual results for several matrix decomposition methods.

Matrix Decompositions
mathematics
linear algebra
visualization

Create matrices and compare how common decompositions reveal structure from different angles.

ROC, KS, cumulative gains, and log-loss charts for a selected classification model.

Model Evaluation Curve
machine learning
model evaluation
credit risk

Visualize classification model performance using ROC, KS, cumulative gains, and log-loss curves.

A credit-model performance dashboard with permutation importance, metric diagnostics, and SAGE decomposition.

Model Performance Decomposition
machine learning
explainability
credit risk

Decompose credit-model performance by variable using permutation and SAGE, with diagnostics for AUC, Gini, log-loss, and KS.

Nine multiclass decision maps comparing predicted regions and confidence across classifiers.

Multiclass Decision Regions
machine learning
classification
visualization
runtime-server

Compare how nine multiclass classifiers divide a two-dimensional space and express confidence.

Pokémon arranged in two-dimensional PCA, t-SNE, and UMAP projections.

Pokémon Dimensionality Reduction
machine learning
visualization
dimensionality reduction
runtime-server

Explore how PCA, t-SNE, and UMAP arrange Pokémon using stats, types, and egg groups.

Score distributions, an ROC curve, a confusion matrix, and threshold metrics.

ROC Curve
machine learning
classification
statistics

Explore how score distributions, class balance, and thresholds shape ROC curves and classification metrics.

A credit-risk prediction with local SHAP contributions and portfolio diagnostics.

SHAP Explorer
machine learning
explainability
credit risk
runtime-server

Explore how different credit-risk models turn a client profile into a probability of default and SHAP contributions.

A three-dimensional shell above three wireframe projections, with controls for expansion, distance, and translation.

Shell Morphospace
mathematics
simulation
visualization

Explore how three growth parameters create coiled shells, spires, and horn-like forms in 3D and planar projections.

A fitted curve with training data and error charts across smoothing bandwidths.

Underfitting and Overfitting
machine learning
statistics
regression

Adjust a kernel smoother and reveal test data and the true relationship to explore model flexibility and generalization.

PDP, ALE, ICE, and observed-distribution charts for a credit-risk predictor.

Variable Effects
machine learning
explainability
credit risk

Compare how credit-risk models respond to each predictor using PDP, ALE, ICE, and the observed distribution.

Credit-risk bins with Weight of Evidence and Information Value charts.

WoE Binning Lab
machine learning
credit risk
statistics

Explore how risk3r binning parameters transform credit-risk predictors into Weight of Evidence and Information Value.
No matching items