test_iris.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Testing package with Iris dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import python_som\n",
    "# Import libraries for plotting results\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "# Import NumPy and Pandas for storing data\n",
    "import numpy as np\n",
    "import seaborn as sns\n",
    "\n",
    "import python_som\n",
    "\n",
    "# Import timing tools\n",
    "import time"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading dataset w/ Seaborn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load Iris dataset and columns of features and labels\n",
    "iris = sns.load_dataset(\"iris\")\n",
    "target = iris.iloc[:, -1].to_numpy()\n",
    "iris = iris.iloc[:, :-1].to_numpy()\n",
    "# Transform labels into numeric codes for plotting\n",
    "tg = np.zeros(len(target), dtype=int)\n",
    "tg[target == \"setosa\"] = 0\n",
    "tg[target == \"versicolor\"] = 1\n",
    "tg[target == \"virginica\"] = 2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## SOM Instantiation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on function __init__ in module python_som:\n",
      "\n",
      "__init__(self, x: int | None, y: int | None, input_len: int, learning_rate: float = 0.5, learning_rate_decay: Callable[[float, int, int], float] = <function _asymptotic_decay at 0x72af5d25b1a0>, neighborhood_radius: float = 1.0, neighborhood_radius_decay: Callable[[float, int, int], float] = <function _asymptotic_decay at 0x72af5d25b1a0>, neighborhood_function: str = 'gaussian', distance_function: Callable[[numpy.ndarray, numpy.ndarray], numpy.ndarray] = <function _euclidean_distance at 0x72af53f94fe0>, cyclic_x: bool = False, cyclic_y: bool = False, random_seed: int | None = None, data: numpy.ndarray | pandas.core.frame.DataFrame | list | None = None) -> None\n",
      "    Constructor for the self-organizing map class.\n",
      "\n",
      "    :param x: int or NoneType: X dimension of the self-organizing map, i.e.,\n",
      "        number of rows of the matrix of weights.\n",
      "        x should be larger than 0.\n",
      "        If x is None and 'data' is provided in kwargs, its value will be automatically\n",
      "        selected using PCA of 'data'. Either x or y should be different than None.\n",
      "    :param y: int or NoneType: Y dimension of the self-organizing map, i.e.,\n",
      "        number of columns of the matrix of weights.\n",
      "        y should be larger than 0.\n",
      "        If y is None and 'data' is provided in kwargs, its value will be automatically\n",
      "        selected using PCA of 'data'. Either x or y should be different than None.\n",
      "    :param input_len: int: Number of features of the training dataset, i.e.,\n",
      "        number of elements of each node of the network.\n",
      "    :param learning_rate: float: Initial learning rate for the training process.\n",
      "        Should be a positive floating point value.\n",
      "        Defaults to 0.5.\n",
      "        Note: The value of the learning_rate is irrelevant for the 'batch' training mode.\n",
      "    :param learning_rate_decay: function: Decay function for the learning_rate variable.\n",
      "        May be a predefined one from this package, or a custom function, with the same\n",
      "        parameters and return type. Defaults to _asymptotic_decay.\n",
      "    :param neighborhood_radius: float: Initial neighborhood radius for the training process.\n",
      "        Defaults to 1.\n",
      "    :param neighborhood_radius_decay: function: Decay function for the neighborhood_radius\n",
      "        variable. May be a predefined one from this package, or a custom function, with the\n",
      "        same parameters and return type. Defaults to _asymptotic_decay\n",
      "    :param neighborhood_function: str: Neighborhood function name for the training process.\n",
      "        May be either 'gaussian' or 'bubble'.\n",
      "    :param distance_function: function: Function for calculating distances/dissimilarities\n",
      "        between models of the network.\n",
      "        May be a predefined one from this package, or a custom function, with the same\n",
      "        parameters and return type. Defaults to _euclidean_distance.\n",
      "    :param cyclic_x: bool: Boolean value activate/deactivate cyclic arrays in the x direction,\n",
      "        i.e, between the first and last rows of the weight matrix.\n",
      "        Defaults to False.\n",
      "    :param cyclic_y: bool: Boolean value activate/deactivate cyclic arrays in the y direction,\n",
      "        i.e, between the first and last columns of the weight matrix.\n",
      "        Defaults to False.\n",
      "    :param random_seed: int or None: Seed for NumPy random value generator. Defaults to None.\n",
      "    :param data: array-like: dataset for performing PCA.\n",
      "        Required when either x or y is None, for determining map size.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "help(python_som.SOM.__init__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Instantiate SOM from  python_som\n",
    "# Selecting shape automatically (providing dataset for constructor)\n",
    "# Using default decay and distance functions\n",
    "# Using gaussian neighborhood function\n",
    "# Using cyclic arrays in the vertical and horizontal directions\n",
    "som = python_som.SOM(x=20, y=None, input_len=iris.shape[1], learning_rate=0.5, neighborhood_radius=1.0,\n",
    "                     neighborhood_function='gaussian', cyclic_x=True, cyclic_y=True, data=iris)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Instantiate SOM from  python_som\n",
    "# # With fixed shape\n",
    "# # Using default decay and distance functions\n",
    "# # Using gaussian neighborhood function\n",
    "# # Not using cyclic arrays\n",
    "# som = python_som.SOM(x=8, y=8, input_len=iris.shape[1], learning_rate=0.5, neighborhood_radius=1.0,\n",
    "#                      neighborhood_function='gaussian', cyclic_x=False, cyclic_y=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Weight initialization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on method weight_initialization in module python_som:\n",
      "\n",
      "weight_initialization(mode: str = 'random', **kwargs: numpy.ndarray | pandas.core.frame.DataFrame | list | str | int) -> None method of python_som.SOM instance\n",
      "    Function for weight initialization of the self-organizing map.\n",
      "\n",
      "    Calls other methods for each initialization mode.\n",
      "\n",
      "    :param mode: str: Initialization mode. May be either 'random', 'linear', or 'sample'.\n",
      "        Note: Each initialization method may require multiple additional arguments in kwargs.\n",
      "    :param kwargs:\n",
      "        For 'random' initialization mode, 'sample_mode': str may be provided to determine\n",
      "        the sampling mode. 'sample_mode' may be either 'standard_normal' (default) or 'uniform'\n",
      "        For 'random' and 'sample' modes, 'random_seed': int may be provided for the random\n",
      "        value generator. For 'sample' and 'linear' modes, 'data': array-like must be provided\n",
      "        for sampling/PCA.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "help(som.weight_initialization)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Initialize weights of the SOM with random initialization, from uniform distribution\n",
    "# som.weight_initialization(mode='random', sample_mode='uniform')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Initialize weights of the SOM with sampling from dataset\n",
    "# som.weight_initialization(mode='sample', data=iris)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Initialize weights of the SOM with linear initialization\n",
    "som.weight_initialization(mode='linear', data=iris)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on method train in module python_som:\n",
      "\n",
      "train(data: numpy.ndarray | pandas.core.frame.DataFrame | list, n_iteration: int | None = None, mode: str = 'random', verbose: bool = False) -> float method of python_som.SOM instance\n",
      "    Trains the self-organizing map, with the dataset 'data', and a certain number of iterations.\n",
      "\n",
      "    :param data: array-like: Dataset for training.\n",
      "    :param n_iteration: int or None: Number of iterations of training.\n",
      "        If None, defaults to 1000 * len(data) for stepwise training modes,\n",
      "        or 10 * len(data) for batch training mode.\n",
      "    :param mode: str: Training mode name. May be either 'random', 'sequential', or 'batch'.\n",
      "        For 'batch' mode, a much smaller number of iterations is needed,\n",
      "        but a higher computation power is required for each individual iteration.\n",
      "    :param verbose: bool: Activate to print useful information to the terminal/console, e.g.,\n",
      "        the progress of the training process\n",
      "    :return: float: Quantization error after training\n",
      "\n"
     ]
    }
   ],
   "source": [
    "help(som.train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Training with stepwise algorithm, with random sampling from the dataset\n",
    "# t_start = time.time()\n",
    "# q_error = som.train(iris, 1000 * len(iris), mode='random', verbose=False)\n",
    "# t_end = time.time()\n",
    "# print(\"Quantization error:\", q_error)\n",
    "# print(\"Execution time:\", t_end-t_start, \"s\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Training with stepwise algorithm, with sequential sampling from the dataset\n",
    "# t_start = time.time()\n",
    "# q_error = som.train(iris, 1000 * len(iris), mode='sequential', verbose=False)\n",
    "# t_end = time.time()\n",
    "# print(\"Quantization error:\", q_error)\n",
    "# print(\"Execution time:\", t_end-t_start, \"s\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Quantization error: 0.11333343581048994\n",
      "Execution time: 6.359957695007324 s\n"
     ]
    }
   ],
   "source": [
    "# Training with batch algorithm\n",
    "t_start = time.time()\n",
    "q_error = som.train(iris, len(iris), mode='batch', verbose=False)\n",
    "t_end = time.time()\n",
    "print(\"Quantization error:\", q_error)\n",
    "print(\"Execution time:\", t_end-t_start, \"s\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Plotting U-matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on method distance_matrix in module python_som:\n",
      "\n",
      "distance_matrix(normalize: bool = False) -> numpy.ndarray method of python_som.SOM instance\n",
      "    Calculates U-matrix of the current state of the network,\n",
      "    i.e., the matrix of distances between each node and its neighbors.\n",
      "    Has support for cyclic arrays\n",
      "\n",
      "    :param normalize: bool: Activate to normalize the U-matrix between 0 and 1.\n",
      "        Defaults to False.\n",
      "    :return: np.ndarray: U-matrix of the current state of the network.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "help(som.distance_matrix)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Calculating distance matrix for plotting\n",
    "umatrix = som.distance_matrix().T\n",
    "\n",
    "# Plotting U-matrix with seaborn/matplotlib\n",
    "plt.figure(figsize=som.get_shape())\n",
    "plt.pcolor(umatrix, cmap='bone_r')\n",
    "\n",
    "markers = ['o', 's', 'D']\n",
    "colors = ['C0', 'C1', 'C2']\n",
    "for cnt, xx in enumerate(iris):\n",
    "    w = som.winner(xx)  # getting the winner\n",
    "    plt.plot(w[0] + .5, w[1] + .5, markers[tg[cnt]], markerfacecolor='None',\n",
    "             markeredgecolor=colors[tg[cnt]], markersize=12, markeredgewidth=2)\n",
    "plt.axis([0, som.get_shape()[0], 0, som.get_shape()[1]])\n",
    "plt.show()\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "torch",
   "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": "3.12.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}