Skip to content

Machine Learning,
Made Visible.

A hands-on way to learn ml, experiment with models, mess with the parameters, and watch the results change in real time.

Launch VisualizerBrowse algorithms
Decision-boundary diagram separating two classes of data points with a single model boundary
Schematic decision-boundary illustration. No measured accuracy is shown.

Live product preview

Open the visualizer

Confluence  /  Logistic Regression

DECISION BOUNDARY · C = 1.0

● CLASS 0 ■ CLASS 1

Run summary

Example summary — open the visualizer to run it live and see your own numbers.

ALGORITHM
logistic-regression
DATASET
breast-cancer
SAMPLES
300
ACCURACY
0.92
F1-SCORE
0.94

LEARNING CURVE

Try it live

SCIKIT-LEARN · EXAMPLE PREVIEW · 300 ROWS

Features

Built for understanding.

Not another toy demo. Confluence runs real algorithms with real computation, wrapped in an interface designed for exploration.

01

Interactive visualization

See decision boundaries, loss curves, classifications, and model behavior update as you change hyperparameters.

02

Real data

Explore 25 real datasets — from Iris and Breast Cancer to Titanic and California Housing — instead of toy-only examples.

03

First-principles learning

Understand what each algorithm is actually optimizing, with theory, formulas, and visual intuition side by side.

04

Explanations

Turn model output into something readable: feature contributions, decision paths, and plain-language prediction breakdowns.

05

Algorithm comparison

Race algorithms on the same dataset, overlay their boundaries, and see exactly where — and why — they diverge.

06

Learning notes

Connect visual behavior with the math through a curated roadmap from linear algebra to model evaluation.

Schematic transformation from raw data points to a model boundary to prediction regions
Conceptual diagram only. No measured values are shown.

Data → Model → Behavior

Algorithms

Algorithms, without the black box.

Every model runs for real, server-side. Filter by family or search to find the one you want to take apart.

Showing 10 of 38

38 algorithms · 4 families

Browse the full encyclopedia

Datasets

Real data.
Real behavior.

Every dataset ships with its full shape — rows, features, task, and target — so you know exactly what the model is learning from.

Explore datasets in the visualizer
Schematic data-topology diagram with clustered points, one outlier, and one selected target
Conceptual data distribution only. Point counts are illustrative, not dataset statistics.

  • Breast Cancer

    Classification

    569 rows · 30 features · target: diagnosis

  • Digits

    Classification

    1,797 rows · 64 features · target: digit

  • Titanic

    Classification

    1,309 rows · 6 features · target: survived

  • Heart Disease

    Classification

    1,025 rows · 7 features · target: target

  • Penguins

    Classification

    342 rows · 2 features · target: species

  • Iris

    Classification

    150 rows · 4 features · target: species

+ 19 more in the visualizer

Resources

Learn the ideas behind the visuals.

  • Machine Learning
  • Linear Algebra
  • Probability
  • Statistics
  • Model Evaluation
  • Feature Engineering

Architecture

From dataset to understanding.

Follow an experiment from configuration through real model computation to visual insight. Documented like infrastructure, not marketing.

REACT · FASTAPI · SCIKIT-LEARN · WEBSOCKET

  1. 01

    Dataset

    25 datasets

    Real and synthetic datasets, projected to 2D for the canvas.

  2. 02

    Preprocessing

    pandas · sklearn

    Scaling, encoding, and train/test splits before any fitting.

  3. 03

    Model Engine

    38 algorithms

    Genuine scikit-learn estimators fit server-side on your data.

  4. 04

    Prediction / Metrics

    numpy · scipy

    Accuracy, ROC-AUC, R², and silhouette computed from real output.

  5. 05

    Visualization

    canvas · three.js

    Decision boundaries, training curves, and 3D surfaces rendered live.

  6. 06

    Explanation

    first principles

    Feature contributions and decision paths in plain language.

Understand the model. Don’t just run it.

Open the visualizer and see the computation happen step by step.

Launch visualizerBrowse algorithms
  • Real computation
  • Open source
  • First principles