k_mean_clustering_from_scratch.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "I7a8MCBdd3hJ"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "from sklearn.datasets import make_blobs\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.cluster import KMeans\n",
        "from sklearn.metrics import silhouette_score\n",
        "import random"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "centroids = [(-5,-5),(5,5), (-2.5, -2.5)]\n",
        "cluster_std = [1,1,1]"
      ],
      "metadata": {
        "id": "Ztd7YY9BeM2r"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "X,y = make_blobs(n_samples=100,centers=centroids,cluster_std=cluster_std, n_features=2, random_state=2)"
      ],
      "metadata": {
        "id": "cZ9vAXL8eVDL"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "class KMeans:\n",
        "  def __init__(self, n_clusters, max_iter):\n",
        "    self.n_clusters = n_clusters\n",
        "    self.max_iter = max_iter\n",
        "    self.centroids = None\n",
        "  def fit_predict(self, X):\n",
        "    random_index = random.sample(range(0,X.shape[0]), self.n_clusters)\n",
        "    self.centroids = X[random_index]\n",
        "    for i in range(self.max_iter):\n",
        "      cluster_group = self.assign_clusters(X)\n",
        "      old_centroids = self.centroids\n",
        "      self.centroids = self.move_centroids(X, cluster_group)\n",
        "      if(self.centroids == old_centroids).all():\n",
        "        break\n",
        "    return cluster_group\n",
        "  def assign_clusters(self, X):\n",
        "    cluster_group = []\n",
        "    distances = []\n",
        "    for row in X:\n",
        "      for centroid in self.centroids:\n",
        "        distances.append(np.sqrt(np.dot(row-centroid,row-centroid)))\n",
        "      min_distance = min(distances)\n",
        "      index_pos = distances.index(min_distance)\n",
        "      cluster_group.append(index_pos)\n",
        "      distances.clear()\n",
        "    return np.array(cluster_group)\n",
        "  def move_centroids(self, X, cluster_group):\n",
        "    new_centroids = []\n",
        "    cluster_type = np.unique(cluster_group)\n",
        "    for type in cluster_type:\n",
        "      new_centroids.append(np.mean(X[cluster_group == type], axis=0))\n",
        "    return np.array(new_centroids)\n"
      ],
      "metadata": {
        "id": "xts3VPGaecCE"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "km = KMeans(n_clusters=3, max_iter=100)\n",
        "y_means = km.fit_predict(X)\n"
      ],
      "metadata": {
        "id": "v_SW4-taeiIt"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plt.scatter(X[y_means == 0,0], X[y_means == 0,1], color='red')\n",
        "plt.scatter(X[y_means == 1,0], X[y_means == 1,1], color='blue')\n",
        "plt.scatter(X[y_means == 2,0], X[y_means == 2,1], color='green')\n",
        "\n",
        "plt.scatter(km.centroids[:,0], km.centroids[:,1], color='black', marker='x')\n",
        "plt.show()"
      ],
      "metadata": {
        "id": "_Qll4Bngwd7U",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 430
        },
        "outputId": "1e7f85dd-3dea-4adf-8b87-bbbc451004e6"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [],
      "metadata": {
        "id": "9lDIk5o2Hht6"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}