{
  "name": "scikit-learn TSNE constructor arguments, defaults and documented guidance",
  "description": "Parameters of sklearn.manifold.TSNE with the default value and the guidance the official scikit-learn API reference and manifold user guide give for each, compiled from the documentation read on 2026-11-14.",
  "page": "https://casrai.org/guides/sklearn-t-sne",
  "publisher": "CASRAI",
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "dateModified": "2026-11-16T22:03:06Z",
  "temporalCoverage": null,
  "spatialCoverage": null,
  "columns": [
    {
      "name": "Parameter",
      "unit": "",
      "description": ""
    },
    {
      "name": "Default",
      "unit": "",
      "description": ""
    },
    {
      "name": "Documented guidance",
      "unit": "",
      "description": ""
    }
  ],
  "rows": [
    {
      "Parameter": "n_components",
      "Default": "2",
      "Documented guidance": "Dimension of the embedded space; Barnes-Hut requires 3 or fewer"
    },
    {
      "Parameter": "perplexity",
      "Default": "30.0",
      "Documented guidance": "Consider 5-50; must be less than the number of samples; larger/noisier data need larger values"
    },
    {
      "Parameter": "early_exaggeration",
      "Default": "12.0",
      "Documented guidance": "Usually need not be tuned; if too high the KL divergence can increase"
    },
    {
      "Parameter": "learning_rate",
      "Default": "auto",
      "Documented guidance": "auto = max(N / early_exaggeration / 4, 50); other implementations define it 4x smaller"
    },
    {
      "Parameter": "max_iter",
      "Default": "1000",
      "Documented guidance": "Should be at least 250; renamed from n_iter in 1.5"
    },
    {
      "Parameter": "n_iter_without_progress",
      "Default": "300",
      "Documented guidance": "Iterations without progress before optimisation stops"
    },
    {
      "Parameter": "min_grad_norm",
      "Default": "1e-07",
      "Documented guidance": "Optimisation stops when the gradient norm falls below this"
    },
    {
      "Parameter": "metric",
      "Default": "euclidean",
      "Documented guidance": "'precomputed' means X is a distance matrix"
    },
    {
      "Parameter": "init",
      "Default": "pca",
      "Documented guidance": "random, pca or an array; pca cannot be used with precomputed distances; default since 1.2"
    },
    {
      "Parameter": "method",
      "Default": "barnes_hut",
      "Documented guidance": "barnes_hut is O(N log N); exact is O(N^2) and used when nearest-neighbour errors must be under 3%"
    },
    {
      "Parameter": "angle",
      "Default": "0.5",
      "Documented guidance": "Barnes-Hut trade-off; 0.2-0.8 and not very sensitive"
    },
    {
      "Parameter": "random_state",
      "Default": "None",
      "Documented guidance": "Set an integer for a repeatable embedding"
    },
    {
      "Parameter": "n_jobs",
      "Default": "None",
      "Documented guidance": "Parallel jobs for neighbour search; ignored for precomputed or euclidean with exact"
    }
  ],
  "sources": [
    {
      "title": "scikit-learn: sklearn.manifold.TSNE",
      "url": "https://scikit-learn.org/stable/modules/generated/sklearn.manifold.TSNE.html",
      "accessed": "2026-11-14"
    },
    {
      "title": "scikit-learn user guide: Manifold learning",
      "url": "https://scikit-learn.org/stable/modules/manifold.html",
      "accessed": "2026-11-14"
    }
  ]
}