04_clustering_comparison.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load necessary libraries."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "### Basics\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.gridspec as gridspec\n",
    "import seaborn as sns\n",
    "import os\n",
    "\n",
    "### ML packages\n",
    "from sklearn.cluster import KMeans, SpectralClustering, AgglomerativeClustering #Hierarchial Clustering\n",
    "from sklearn.mixture import GaussianMixture\n",
    "import scipy.cluster.hierarchy as sch\n",
    "from scipy.cluster.hierarchy import dendrogram, linkage\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "from collections import defaultdict\n",
    "import umap\n",
    "\n",
    "### Msc\n",
    "import warnings\n",
    "\n",
    "### OOP\n",
    "from ml_utils import ClusterEvaluator"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Load dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('00_authors.csv').rename(columns = {'Unnamed: 0': 'Author'}).drop(columns = 'BookID')\n",
    "X = df.copy().drop(['Author'], axis=1)\n",
    "X = X.to_numpy() # change pd.DataFrame to np.ndarray\n",
    "authors = df['Author'].values  # n_samples-length array"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can also assign a hyperparameter $K=4$.  This represents the number of clusters (which we already know; there are 4 authors). Later on we will be hyperparameter tuning for some $K$ based on stability!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "K_ = 4"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let us also define a dictionary to store our results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "accuracy_dict = {}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Clustering Methods"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Kmeans++"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The mapping based on mode = {0: 'Austen', 1: 'Shakespeare', 2: 'London', 3: 'London'}\n",
      "The accuracy of Kmeans++ = 0.9096313912009513\n"
     ]
    }
   ],
   "source": [
    "kmeans = KMeans(n_clusters=K_, init='k-means++', n_init=10, max_iter=300) # K-means++ initialization \n",
    "kmeans.fit(X)\n",
    "y_kmeans = kmeans.predict(X)\n",
    "centers_pp = kmeans.cluster_centers_\n",
    "\n",
    "accuracy, mapping = ClusterEvaluator(y_kmeans,df['Author'].values).accuracy\n",
    "print(f'The mapping based on mode = {mapping}')\n",
    "print(f'The accuracy of Kmeans++ = {accuracy}')\n",
    "accuracy_dict['kmeans++'] = accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Gaussian Mixture Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The mapping based on mode = {0: 'Austen', 1: 'Shakespeare', 2: 'London', 3: 'Milton'}\n",
      "The accuracy of Kmeans++ = 0.9512485136741974\n"
     ]
    }
   ],
   "source": [
    "gmm = GaussianMixture(n_components=K_)\n",
    "gmm.fit(X)\n",
    "y_gmm = gmm.predict(X)\n",
    "\n",
    "accuracy, mapping = ClusterEvaluator(y_gmm,df['Author'].values).accuracy\n",
    "print(f'The mapping based on mode = {mapping}')\n",
    "print(f'The accuracy of Kmeans++ = {accuracy}')\n",
    "accuracy_dict['gmm'] = accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Spectral Clustering"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Spectral clustering already had a dimensional reduction by design, i.e., spectral clustering is essentially spectral embedding followed by kmeans!  Therefore, we expect this 'raw' method to perform best compared to the other methods without any form of dimensionality reduction!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The mapping based on mode = {0: 'Shakespeare', 1: 'Austen', 2: 'London', 3: 'Milton'}\n",
      "The accuracy of Kmeans++ = 0.9881093935790726\n"
     ]
    }
   ],
   "source": [
    "spectral = SpectralClustering(n_clusters=K_, affinity='nearest_neighbors')\n",
    "y_spectral = spectral.fit_predict(X)\n",
    "\n",
    "accuracy, mapping = ClusterEvaluator(y_spectral,df['Author'].values).accuracy\n",
    "print(f'The mapping based on mode = {mapping}')\n",
    "print(f'The accuracy of Kmeans++ = {accuracy}')\n",
    "accuracy_dict['spectral clustering'] = accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Hierarchial Clustering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Establish all possible combinations for linkage and metric!\n",
    "methods = []\n",
    "linkage_methods = ['ward', 'average', 'complete', 'single']\n",
    "metrics = ['euclidean', 'manhattan', 'cosine']\n",
    "\n",
    "for linkage in linkage_methods:\n",
    "    for metric in metrics:\n",
    "        if linkage == 'ward' and metric != 'euclidean':\n",
    "            continue  # ward only supports euclidean\n",
    "        methods.append((linkage, metric))\n",
    "\n",
    "# Loop through all combinations and determine best accuracy!\n",
    "hierarchical_accuracy_dict = {}\n",
    "vals = df['Author'].values\n",
    "for linkage, metric in methods:\n",
    "    hierarchical = AgglomerativeClustering(n_clusters=K_, linkage=linkage, metric=metric)\n",
    "    y_hierarchical = hierarchical.fit_predict(X)  # Fit and predict in one step\n",
    "    accuracy, mapping = ClusterEvaluator(y_hierarchical,vals).accuracy # Use OOP code\n",
    "    # print(f'The mapping based on mode = {mapping}')\n",
    "    # print(f'The accuracy hierarchial clustering with {linkage} and {metric} = {accuracy}')\n",
    "    hierarchical_accuracy_dict[f'{linkage}-{metric}'] = accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "keys = list(hierarchical_accuracy_dict.keys())\n",
    "vals = list(hierarchical_accuracy_dict.values())\n",
    "colors = plt.colormaps.get_cmap('tab10').colors\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(20, 5))\n",
    "bars = ax.bar(keys, vals, color=colors)\n",
    "ax.bar_label(bars, fmt='%.3f', padding=3)\n",
    "\n",
    "ax.set_ylabel('Accuracy')\n",
    "ax.set_title('Accuracy - Metric & Linkage')\n",
    "plt.tight_layout()\n",
    "plt.savefig('Media/viz/04/04_hier_accuracy_across_metric_linkage')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lets store the highest accuracy params to compare across other methods - ward + euclidean."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "accuracy_dict['hierarchial (ward-euclidean)'] = hierarchical_accuracy_dict['ward-euclidean']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "keys = list(accuracy_dict.keys())\n",
    "vals = list(accuracy_dict.values())\n",
    "\n",
    "colors = ['#bcbd22', '#17becf', '#f7b6d2', '#c49c94']\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10, 5))\n",
    "bars = ax.bar(keys, vals, color=colors)\n",
    "ax.bar_label(bars, fmt='%.3f', padding=3)\n",
    "\n",
    "ax.set_ylabel('Accuracy')\n",
    "ax.set_title('Accuracy - Metric & Linkage')\n",
    "plt.tight_layout()\n",
    "plt.savefig('Media/viz/04/04_accuracy_across_methods')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Clearly, spectral clustering performs the best because it has a build in dimensional reduction! Therefore, now let us apply UMAP to all these methods and recompare."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Dimensional Reduction + Clustering Methods"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Use UMAP as dimensionality reduction for all methods (except spectral clustering as this already uses spectral embedding so we will not repeat this)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "warnings.filterwarnings(\"ignore\", category=UserWarning, module=\"umap\") # Suppress the specific UMAP warning on parralelism\n",
    "umap_model = umap.UMAP(n_neighbors=10, min_dist=0.3)\n",
    "X_umap = umap_model.fit_transform(X)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Store results in the following dictionary to compare across methods."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "umap_accuracy_dict = {}\n",
    "umap_accuracy_dict['spectral clustering'] = accuracy_dict['spectral clustering']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Kmeans++"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The mapping based on mode = {0: 'Austen', 1: 'London', 2: 'Shakespeare', 3: 'Milton'}\n",
      "The accuracy of Kmeans++ = 0.9916765755053508\n"
     ]
    }
   ],
   "source": [
    "# Run K-means++ on UMAP-reduced data\n",
    "kmeans = KMeans(n_clusters=K_, init='k-means++', n_init=10, max_iter=300)\n",
    "kmeans.fit(X_umap)\n",
    "y_kmeans = kmeans.predict(X_umap)\n",
    "centers_pp = kmeans.cluster_centers_\n",
    "\n",
    "accuracy, mapping = ClusterEvaluator(y_kmeans,df['Author'].values).accuracy\n",
    "print(f'The mapping based on mode = {mapping}')\n",
    "print(f'The accuracy of Kmeans++ = {accuracy}')\n",
    "umap_accuracy_dict['kmeans++'] = accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Gaussian Mixture Models"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The mapping based on mode = {0: 'Shakespeare', 1: 'London', 2: 'Austen', 3: 'Milton'}\n",
      "The accuracy of Kmeans++ = 0.9916765755053508\n"
     ]
    }
   ],
   "source": [
    "gmm = GaussianMixture(n_components=K_)\n",
    "gmm.fit(X_umap)\n",
    "y_gmm = gmm.predict(X_umap)\n",
    "\n",
    "accuracy, mapping = ClusterEvaluator(y_gmm,df['Author'].values).accuracy\n",
    "print(f'The mapping based on mode = {mapping}')\n",
    "print(f'The accuracy of Kmeans++ = {accuracy}')\n",
    "umap_accuracy_dict['gmm'] = accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Hierarchial Clustering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The mapping based on mode = {0: 'Austen', 1: 'London', 2: 'Shakespeare', 3: 'Milton'}\n",
      "The accuracy hierarchial clustering with ward and euclidean = 0.9916765755053508\n"
     ]
    }
   ],
   "source": [
    "linkage = 'ward'\n",
    "metric = 'euclidean'\n",
    "hierarchical = AgglomerativeClustering(n_clusters=K_, linkage=linkage, metric=metric)\n",
    "y_hierarchical = hierarchical.fit_predict(X_umap)  # Fit and predict in one step\n",
    "accuracy, mapping = ClusterEvaluator(y_hierarchical,df['Author'].values).accuracy # Use OOP code\n",
    "print(f'The mapping based on mode = {mapping}')\n",
    "print(f'The accuracy hierarchial clustering with {linkage} and {metric} = {accuracy}')\n",
    "umap_accuracy_dict[f'hierarchial ({linkage}-{metric})'] = accuracy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Comparison"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As we can see, all these methods (with the exception of spectral clustering as this did not use UMAP) converge to the same accuracy.  This is because after applying UMAP we get very clearly defined clusters so there is not much/any room for these methods to diverge in their clustering assignments!  Therefore they will yield the same accuracy; this makes sense!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "keys = list(umap_accuracy_dict.keys())\n",
    "vals = list(umap_accuracy_dict.values())\n",
    "\n",
    "colors = ['#9467bd', '#8c564b', '#e377c2', '#7f7f7f']\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10, 5))\n",
    "bars = ax.bar(keys, vals, color=colors)\n",
    "ax.bar_label(bars, fmt='%.3f', padding=3)\n",
    "\n",
    "ax.set_ylabel('Accuracy')\n",
    "ax.set_title('Accuracy w/ UMAP - Metric & Linkage')\n",
    "plt.tight_layout()\n",
    "plt.savefig('Media/viz/04/04_accuracy_across_methods_with_umap')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "base",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}