crossref_analysis_3_lda_hdbscan.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "fucktheseabstracts.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "qfCZZk3ghWMD",
        "colab_type": "code",
        "outputId": "71c87bc8-dd3b-4455-901c-d3be050f5670",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 371
        }
      },
      "source": [
        "! pip install hdbscan"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting hdbscan\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/22/2f/2423d844072f007a74214c1adc46260e45f034bb1679ccadfbb8a601f647/hdbscan-0.8.26.tar.gz (4.7MB)\n",
            "\u001b[K     |████████████████████████████████| 4.7MB 2.8MB/s \n",
            "\u001b[?25h  Installing build dependencies ... \u001b[?25l\u001b[?25hdone\n",
            "  Getting requirements to build wheel ... \u001b[?25l\u001b[?25hdone\n",
            "    Preparing wheel metadata ... \u001b[?25l\u001b[?25hdone\n",
            "Requirement already satisfied: joblib in /usr/local/lib/python3.6/dist-packages (from hdbscan) (0.15.1)\n",
            "Requirement already satisfied: scipy>=0.9 in /usr/local/lib/python3.6/dist-packages (from hdbscan) (1.4.1)\n",
            "Requirement already satisfied: cython>=0.27 in /usr/local/lib/python3.6/dist-packages (from hdbscan) (0.29.18)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from hdbscan) (1.12.0)\n",
            "Requirement already satisfied: scikit-learn>=0.17 in /usr/local/lib/python3.6/dist-packages (from hdbscan) (0.22.2.post1)\n",
            "Requirement already satisfied: numpy>=1.16.0 in /usr/local/lib/python3.6/dist-packages (from hdbscan) (1.18.4)\n",
            "Building wheels for collected packages: hdbscan\n",
            "  Building wheel for hdbscan (PEP 517) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for hdbscan: filename=hdbscan-0.8.26-cp36-cp36m-linux_x86_64.whl size=2308907 sha256=8c150d8aed8be041deb17ff284540a606759a74473adabdbf85e6e511ea54cfd\n",
            "  Stored in directory: /root/.cache/pip/wheels/82/38/41/372f034d8abd271ef7787a681e0a47fc05d472683a7eb088ed\n",
            "Successfully built hdbscan\n",
            "Installing collected packages: hdbscan\n",
            "Successfully installed hdbscan-0.8.26\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QFE4ToOjhVmN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import hdbscan"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FAfxhCzihbbj",
        "colab_type": "code",
        "outputId": "70b02b15-eb51-476c-fcf9-576122299f32",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 72
        }
      },
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/statsmodels/tools/_testing.py:19: FutureWarning: pandas.util.testing is deprecated. Use the functions in the public API at pandas.testing instead.\n",
            "  import pandas.util.testing as tm\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5YEvMO6tpv0E",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from scipy.sparse import load_npz\n",
        "\n",
        "features = load_npz('/content/features_3cat_geo.npz')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "STySO7-SbUFY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "labels = np.load('labels_3cat_geo.npy', allow_pickle=True)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WU6X_O42SBDj",
        "colab_type": "code",
        "outputId": "55f12208-6ac7-4d1d-f7a0-13925c1014df",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "features.shape, labels.shape"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((1480, 160444), (1480,))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6nBAhqLbRsMq",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.discriminant_analysis import LinearDiscriminantAnalysis as LDA"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FHBnKLtenDpN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "lda = LDA(n_components=2)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3BKWKKXinO1H",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "X_train, X_test, y_train, y_test = train_test_split(features, labels, train_size=.9, test_size=.1,\n",
        "                                                    random_state=1337, stratify=labels)\n",
        "\n",
        "pos = lda.fit_transform(X_train.toarray(), y_train)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-PKvZlalXHRG",
        "colab_type": "code",
        "outputId": "15ac55fa-4e7f-4686-83ca-d074cb28cfb9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "pos.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(1332, 2)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 179
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cLQ_rbpvbayX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "encoding = {subj: i for i, subj in enumerate(np.unique(labels))}\n",
        "labels_encoded = [encoding[label] for label in labels]"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ejVyUZ5Ceby3",
        "colab_type": "code",
        "outputId": "1b32b421-6ab5-4344-d2a0-fcca556dda3a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 238
        }
      },
      "source": [
        "color_palette = sns.color_palette('deep', np.unique(labels).shape[0])\n",
        "cluster_colors = [color_palette[x] for x in labels_encoded]\n",
        "plt.scatter(*features.T, s=50, linewidth=0, c=cluster_colors, alpha=0.5)\n",
        "plt.title('True labels')\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "error",
          "ename": "TypeError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-212-79871d93a955>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mcolor_palette\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msns\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcolor_palette\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'deep'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0munique\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlabels\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[0mcluster_colors\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mcolor_palette\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mx\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mlabels_encoded\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mscatter\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0mfeatures\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mT\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0ms\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m50\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlinewidth\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mc\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcluster_colors\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0malpha\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m0.5\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtitle\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'True labels'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshow\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mTypeError\u001b[0m: scatter() got multiple values for argument 's'"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7xw6HfABpzH5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "clusterer = hdbscan.HDBSCAN(min_cluster_size=15,\n",
        "                            min_samples=1).fit(features)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Xp1exxJ1OveS",
        "colab_type": "code",
        "outputId": "1fd75f02-e37e-4ecd-c491-6d151e2651d4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "clusterer.labels_.max()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "2"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 220
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xedgY2ffjw8H",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "clusterer.condensed_tree_.plot()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vdDhFNsM8OLs",
        "colab_type": "code",
        "outputId": "f4d71852-f3c7-4864-a784-4f9615efbf7a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        }
      },
      "source": [
        "color_palette = sns.color_palette('deep', clusterer.labels_.max() + 1)\n",
        "cluster_colors = [color_palette[x] if x >= 0\n",
        "                  else (0.5, 0.5, 0.5)\n",
        "                  for x in clusterer.labels_]\n",
        "cluster_member_colors = [sns.desaturate(x, p) for x, p in\n",
        "                         zip(cluster_colors, clusterer.probabilities_)]\n",
        "plt.scatter(*pos.T, s=50, linewidth=0, c=cluster_member_colors, alpha=0.5)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.collections.PathCollection at 0x7f98fc72f160>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 184
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "p8mKNVSOIpNd",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.cluster import KMeans\n",
        "\n",
        "km = KMeans(n_clusters=np.unique(labels).shape[0], verbose=True)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qxLMyYhQoacc",
        "colab_type": "code",
        "outputId": "92bc3e69-e0be-41ae-e525-2d9b86c69121",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "km.fit(features)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Initialization complete\n",
            "Iteration  0, inertia 2846.805\n",
            "Iteration  1, inertia 1437.932\n",
            "Iteration  2, inertia 1437.736\n",
            "Iteration  3, inertia 1437.646\n",
            "Iteration  4, inertia 1437.632\n",
            "Converged at iteration 4: center shift 0.000000e+00 within tolerance 6.069252e-10\n",
            "Initialization complete\n",
            "Iteration  0, inertia 2849.501\n",
            "Iteration  1, inertia 1436.715\n",
            "Iteration  2, inertia 1436.112\n",
            "Iteration  3, inertia 1435.818\n",
            "Iteration  4, inertia 1435.763\n",
            "Iteration  5, inertia 1435.744\n",
            "Iteration  6, inertia 1435.740\n",
            "Iteration  7, inertia 1435.737\n",
            "Converged at iteration 7: center shift 0.000000e+00 within tolerance 6.069252e-10\n",
            "Initialization complete\n",
            "Iteration  0, inertia 2845.157\n",
            "Iteration  1, inertia 1438.209\n",
            "Iteration  2, inertia 1437.653\n",
            "Iteration  3, inertia 1436.889\n",
            "Iteration  4, inertia 1436.322\n",
            "Iteration  5, inertia 1436.060\n",
            "Iteration  6, inertia 1435.950\n",
            "Iteration  7, inertia 1435.902\n",
            "Iteration  8, inertia 1435.891\n",
            "Iteration  9, inertia 1435.879\n",
            "Iteration 10, inertia 1435.873\n",
            "Iteration 11, inertia 1435.868\n",
            "Iteration 12, inertia 1435.861\n",
            "Iteration 13, inertia 1435.852\n",
            "Iteration 14, inertia 1435.844\n",
            "Converged at iteration 14: center shift 0.000000e+00 within tolerance 6.069252e-10\n",
            "Initialization complete\n",
            "Iteration  0, inertia 2846.247\n",
            "Iteration  1, inertia 1438.071\n",
            "Iteration  2, inertia 1437.827\n",
            "Iteration  3, inertia 1437.563\n",
            "Iteration  4, inertia 1437.250\n",
            "Iteration  5, inertia 1436.888\n",
            "Iteration  6, inertia 1436.563\n",
            "Iteration  7, inertia 1436.319\n",
            "Iteration  8, inertia 1436.182\n",
            "Iteration  9, inertia 1436.078\n",
            "Iteration 10, inertia 1436.021\n",
            "Iteration 11, inertia 1435.983\n",
            "Iteration 12, inertia 1435.967\n",
            "Iteration 13, inertia 1435.952\n",
            "Iteration 14, inertia 1435.946\n",
            "Iteration 15, inertia 1435.943\n",
            "Iteration 16, inertia 1435.940\n",
            "Converged at iteration 16: center shift 0.000000e+00 within tolerance 6.069252e-10\n",
            "Initialization complete\n",
            "Iteration  0, inertia 2845.679\n",
            "Iteration  1, inertia 1438.277\n",
            "Iteration  2, inertia 1438.232\n",
            "Iteration  3, inertia 1438.171\n",
            "Iteration  4, inertia 1438.129\n",
            "Iteration  5, inertia 1438.113\n",
            "Iteration  6, inertia 1438.103\n",
            "Iteration  7, inertia 1438.088\n",
            "Iteration  8, inertia 1438.073\n",
            "Iteration  9, inertia 1438.068\n",
            "Iteration 10, inertia 1438.053\n",
            "Converged at iteration 10: center shift 0.000000e+00 within tolerance 6.069252e-10\n",
            "Initialization complete\n",
            "Iteration  0, inertia 2833.993\n",
            "Iteration  1, inertia 1437.864\n",
            "Iteration  2, inertia 1437.639\n",
            "Iteration  3, inertia 1437.365\n",
            "Iteration  4, inertia 1437.168\n",
            "Iteration  5, inertia 1436.992\n",
            "Iteration  6, inertia 1436.779\n",
            "Iteration  7, inertia 1436.571\n",
            "Iteration  8, inertia 1436.499\n",
            "Iteration  9, inertia 1436.456\n",
            "Iteration 10, inertia 1436.436\n",
            "Iteration 11, inertia 1436.416\n",
            "Iteration 12, inertia 1436.401\n",
            "Iteration 13, inertia 1436.394\n",
            "Iteration 14, inertia 1436.390\n",
            "Iteration 15, inertia 1436.386\n",
            "Iteration 16, inertia 1436.382\n",
            "Converged at iteration 16: center shift 0.000000e+00 within tolerance 6.069252e-10\n",
            "Initialization complete\n",
            "Iteration  0, inertia 2841.610\n",
            "Iteration  1, inertia 1438.537\n",
            "Iteration  2, inertia 1438.493\n",
            "Iteration  3, inertia 1438.481\n",
            "Iteration  4, inertia 1438.474\n",
            "Iteration  5, inertia 1438.462\n",
            "Iteration  6, inertia 1438.440\n",
            "Iteration  7, inertia 1438.437\n",
            "Iteration  8, inertia 1438.431\n",
            "Iteration  9, inertia 1438.426\n",
            "Iteration 10, inertia 1438.423\n",
            "Iteration 11, inertia 1438.420\n",
            "Iteration 12, inertia 1438.413\n",
            "Iteration 13, inertia 1438.405\n",
            "Converged at iteration 13: center shift 0.000000e+00 within tolerance 6.069252e-10\n",
            "Initialization complete\n",
            "Iteration  0, inertia 2835.156\n",
            "Iteration  1, inertia 1438.560\n",
            "Iteration  2, inertia 1438.492\n",
            "Iteration  3, inertia 1438.275\n",
            "Iteration  4, inertia 1437.573\n",
            "Iteration  5, inertia 1436.644\n",
            "Iteration  6, inertia 1436.158\n",
            "Iteration  7, inertia 1436.006\n",
            "Iteration  8, inertia 1435.967\n",
            "Iteration  9, inertia 1435.955\n",
            "Iteration 10, inertia 1435.949\n",
            "Iteration 11, inertia 1435.946\n",
            "Converged at iteration 11: center shift 0.000000e+00 within tolerance 6.069252e-10\n",
            "Initialization complete\n",
            "Iteration  0, inertia 2843.277\n",
            "Iteration  1, inertia 1438.369\n",
            "Iteration  2, inertia 1438.219\n",
            "Iteration  3, inertia 1438.099\n",
            "Iteration  4, inertia 1437.911\n",
            "Iteration  5, inertia 1437.400\n",
            "Iteration  6, inertia 1436.753\n",
            "Iteration  7, inertia 1436.296\n",
            "Iteration  8, inertia 1436.119\n",
            "Iteration  9, inertia 1436.007\n",
            "Iteration 10, inertia 1435.934\n",
            "Iteration 11, inertia 1435.893\n",
            "Iteration 12, inertia 1435.848\n",
            "Iteration 13, inertia 1435.824\n",
            "Iteration 14, inertia 1435.785\n",
            "Iteration 15, inertia 1435.753\n",
            "Iteration 16, inertia 1435.719\n",
            "Iteration 17, inertia 1435.704\n",
            "Iteration 18, inertia 1435.695\n",
            "Iteration 19, inertia 1435.689\n",
            "Converged at iteration 19: center shift 0.000000e+00 within tolerance 6.069252e-10\n",
            "Initialization complete\n",
            "Iteration  0, inertia 2827.530\n",
            "Iteration  1, inertia 1437.871\n",
            "Iteration  2, inertia 1437.173\n",
            "Iteration  3, inertia 1436.659\n",
            "Iteration  4, inertia 1436.331\n",
            "Iteration  5, inertia 1436.135\n",
            "Iteration  6, inertia 1436.071\n",
            "Iteration  7, inertia 1436.027\n",
            "Iteration  8, inertia 1436.005\n",
            "Iteration  9, inertia 1435.974\n",
            "Iteration 10, inertia 1435.955\n",
            "Iteration 11, inertia 1435.937\n",
            "Converged at iteration 11: center shift 0.000000e+00 within tolerance 6.069252e-10\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "KMeans(algorithm='auto', copy_x=True, init='k-means++', max_iter=300,\n",
              "       n_clusters=3, n_init=10, n_jobs=None, precompute_distances='auto',\n",
              "       random_state=None, tol=0.0001, verbose=True)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 224
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RmsRhDCPobwx",
        "colab_type": "code",
        "outputId": "39c1429c-dba9-4215-ee66-300a7c45d95f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 87
        }
      },
      "source": [
        "from sklearn import metrics\n",
        "\n",
        "print(\"Homogeneity: %0.3f\" % metrics.homogeneity_score(labels, km.labels_))\n",
        "print(\"Completeness: %0.3f\" % metrics.completeness_score(labels, km.labels_))\n",
        "print(\"V-measure: %0.3f\" % metrics.v_measure_score(labels, km.labels_))\n",
        "print(\"Adjusted Rand-Index: %.3f\"\n",
        "      % metrics.adjusted_rand_score(labels, km.labels_))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Homogeneity: 0.060\n",
            "Completeness: 0.092\n",
            "V-measure: 0.072\n",
            "Adjusted Rand-Index: 0.073\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fln1ZHVTolzW",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "pos_test = lda.predict(X_test.toarray())"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FSgdBwPX7h72",
        "colab_type": "code",
        "outputId": "ed5640e0-e05b-45e6-925f-313f13d81055",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "km.n_clusters"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "3"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 232
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Qz7BNyvK7nHt",
        "colab_type": "code",
        "outputId": "9865b68e-5e06-4561-cfda-6325250b2c4f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 87
        }
      },
      "source": [
        "print(\"Homogeneity: %0.3f\" % metrics.homogeneity_score(y_test, pos_test))\n",
        "print(\"Completeness: %0.3f\" % metrics.completeness_score(y_test, pos_test))\n",
        "print(\"V-measure: %0.3f\" % metrics.v_measure_score(y_test, pos_test))\n",
        "print(\"Adjusted Rand-Index: %.3f\"\n",
        "      % metrics.adjusted_rand_score(y_test, pos_test))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Homogeneity: 0.311\n",
            "Completeness: 0.360\n",
            "V-measure: 0.334\n",
            "Adjusted Rand-Index: 0.288\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PgLQyC2m7wng",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}