20 classic machine-learning and CS algorithms, each implemented from scratch in NumPy (no scikit-learn, no PyTorch in the core algorithm) and paired with an interactive Streamlit + Plotly step-by-step walkthrough — all in one categorized multipage app.
Live demo: algorithm-visualizers.streamlit.app
This repo replaces 20 previously-standalone visualiser repos. Each algorithm keeps its own self-contained package (<algo>/{algorithm,data,visualize}.py); only the Streamlit entry point was merged into a single app with one shared home page and navigation.
uv sync
uv run streamlit run Home.pyOpen the URL printed in your terminal (usually http://localhost:8501). The sidebar navigation groups every algorithm by category; the home page gives a card-based overview of all 20.
| Category | Algorithms |
|---|---|
| Clustering | K-Means · DBSCAN · Gaussian Mixture (EM) |
| Dimensionality reduction | PCA · UMAP · t-SNE |
| Classification & ensembles | Perceptron & Gradient Descent · Support Vector Machine · Random Forest |
| Deep learning building blocks | Backpropagation · Transformer Self-Attention |
| Generative & self-supervised models | Variational Autoencoder · Diffusion Model (DDPM) · Contrastive Learning |
| Graph algorithms | Dijkstra & A* · Minimum Spanning Tree (Kruskal & Prim) |
| Probabilistic methods, state estimation & signal processing | Markov Chain Monte Carlo · Kalman Filter · Particle Filter · Fast Fourier Transform |
| Reinforcement learning | Q-Learning / SARSA |
Every visualiser follows the same convention: a from-scratch NumPy implementation, a step-by-step or frame-by-frame playback control, and an in-app explanation of what's happening and why the algorithm can fail.
algorithm-visualizers/
├── Home.py # Landing page + st.navigation wiring
├── apps/<algo>.py # One Streamlit page per algorithm
├── <algo>/ # Each algorithm's own from-scratch package
│ ├── algorithm.py # Core algorithm, records a Snapshot per step
│ ├── data.py # Synthetic dataset / environment generators
│ └── visualize.py # Plotly figure builders
├── .streamlit/config.toml # Shared theme (indigo/teal, Inter font)
└── pyproject.toml
Every page shares one visual identity: an indigo/teal palette, Inter typeface, bordered card layout for parameter groups and charts, and a matching Plotly theme. .streamlit/config.toml and each page's chart palette are the source of truth — kept identical across every algorithm.