python/notebooks/example_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### This example demonstrates the XCSF supervised learning mechanisms to perform classification\n",
    "Classifiers are composed of neural network conditions and predictions. A softmax layer is used as prediction output and labels are one-hot encoded. Similar to regression, a single dummy action is performed such that [A] = [M]."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from __future__ import annotations\n",
    "\n",
    "import json\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from sklearn.datasets import fetch_openml\n",
    "from sklearn.metrics import ConfusionMatrixDisplay\n",
    "from sklearn.metrics import confusion_matrix, classification_report\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import MinMaxScaler, OneHotEncoder\n",
    "from tqdm import tqdm\n",
    "\n",
    "import xcsf\n",
    "\n",
    "RANDOM_STATE: int = 10\n",
    "np.random.seed(RANDOM_STATE)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load training and test data\n",
    "Loads USPS data from: https://www.openml.org/d/41082 \n",
    "\n",
    "\"Normalized handwritten digits, automatically scanned from envelopes by the U.S. Postal Service. The original scanned digits are binary and of different sizes and orientations; the images here have been deslanted and size normalized, resulting in 16 x 16 grayscale images (Le Cun et al., 1990).\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 256 features, 10 classes, 9298 instances\n",
    "data = fetch_openml(data_id=41082, parser=\"auto\")\n",
    "INPUT_HEIGHT: int = 16\n",
    "INPUT_WIDTH: int = 16\n",
    "INPUT_CHANNELS: int = 1\n",
    "\n",
    "# numpy\n",
    "X = np.asarray(data.data, dtype=np.float64)\n",
    "y = np.asarray(data.target, dtype=np.int16)\n",
    "\n",
    "# scale features [0,1]\n",
    "feature_scaler = MinMaxScaler()\n",
    "X = feature_scaler.fit_transform(X)\n",
    "\n",
    "# USPS labels start at 1\n",
    "y = np.subtract(y, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x200 with 24 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 2))\n",
    "for i in range(24):\n",
    "    plt.subplot(2, 12, i + 1)\n",
    "    plt.imshow(X[i].reshape(INPUT_WIDTH, INPUT_HEIGHT), cmap=\"gray_r\")\n",
    "    plt.title(f\"{y[i]}\")\n",
    "    plt.xticks([])\n",
    "    plt.yticks([])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape = (7531, 256)\n",
      "y_train shape = (7531, 10)\n",
      "X_val shape = (837, 256)\n",
      "y_val shape = (837, 10)\n",
      "X_test shape = (930, 256)\n",
      "y_test shape = (930, 10)\n"
     ]
    }
   ],
   "source": [
    "# one hot encode labels\n",
    "onehot_encoder = OneHotEncoder(sparse_output=False, categories=\"auto\")\n",
    "y = onehot_encoder.fit_transform(y.reshape(-1, 1))\n",
    "\n",
    "# split into training and test sets\n",
    "X_train, X_test, y_train, y_test = train_test_split(\n",
    "    X, y, test_size=0.1, random_state=RANDOM_STATE\n",
    ")\n",
    "\n",
    "# 10% of training for validation\n",
    "X_train, X_val, y_train, y_val = train_test_split(\n",
    "    X_train, y_train, test_size=0.1, random_state=RANDOM_STATE\n",
    ")\n",
    "\n",
    "# get number of input and output variables\n",
    "X_DIM: int = np.shape(X_train)[1]\n",
    "Y_DIM: int = np.shape(y_train)[1]\n",
    "\n",
    "print(f\"X_train shape = {np.shape(X_train)}\")\n",
    "print(f\"y_train shape = {np.shape(y_train)}\")\n",
    "print(f\"X_val shape = {np.shape(X_val)}\")\n",
    "print(f\"y_val shape = {np.shape(y_val)}\")\n",
    "print(f\"X_test shape = {np.shape(X_test)}\")\n",
    "print(f\"y_test shape = {np.shape(y_test)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Initialise XCSF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\n",
      "    \"version\": \"1.3.0\",\n",
      "    \"x_dim\": 256,\n",
      "    \"y_dim\": 10,\n",
      "    \"n_actions\": 1,\n",
      "    \"omp_num_threads\": 12,\n",
      "    \"random_state\": 10,\n",
      "    \"population_file\": \"\",\n",
      "    \"pop_init\": true,\n",
      "    \"max_trials\": 100000,\n",
      "    \"perf_trials\": 1000,\n",
      "    \"pop_size\": 100,\n",
      "    \"loss_func\": \"log\",\n",
      "    \"set_subsumption\": false,\n",
      "    \"theta_sub\": 100,\n",
      "    \"e0\": 0.01,\n",
      "    \"alpha\": 1,\n",
      "    \"nu\": 5,\n",
      "    \"beta\": 0.05,\n",
      "    \"delta\": 0.1,\n",
      "    \"theta_del\": 200,\n",
      "    \"init_fitness\": 0.01,\n",
      "    \"init_error\": 0,\n",
      "    \"m_probation\": 10000,\n",
      "    \"stateful\": true,\n",
      "    \"compaction\": false,\n",
      "    \"ea\": {\n",
      "        \"select_type\": \"roulette\",\n",
      "        \"theta_ea\": 200,\n",
      "        \"lambda\": 2,\n",
      "        \"p_crossover\": 0.8,\n",
      "        \"err_reduc\": 1,\n",
      "        \"fit_reduc\": 0.1,\n",
      "        \"subsumption\": false,\n",
      "        \"pred_reset\": false\n",
      "    },\n",
      "    \"condition\": {\n",
      "        \"type\": \"neural\",\n",
      "        \"args\": {\n",
      "            \"layer_0\": {\n",
      "                \"type\": \"connected\",\n",
      "                \"activation\": \"relu\",\n",
      "                \"n_inputs\": 256,\n",
      "                \"n_init\": 1,\n",
      "                \"evolve_weights\": true,\n",
      "                \"evolve_functions\": false,\n",
      "                \"evolve_connect\": true,\n",
      "                \"evolve_neurons\": true,\n",
      "                \"n_max\": 100,\n",
      "                \"max_neuron_grow\": 1,\n",
      "                \"sgd_weights\": false\n",
      "            },\n",
      "            \"layer_1\": {\n",
      "                \"type\": \"connected\",\n",
      "                \"activation\": \"linear\",\n",
      "                \"n_inputs\": 1,\n",
      "                \"n_init\": 1,\n",
      "                \"evolve_weights\": true,\n",
      "                \"evolve_functions\": false,\n",
      "                \"evolve_connect\": true,\n",
      "                \"evolve_neurons\": false,\n",
      "                \"sgd_weights\": false\n",
      "            }\n",
      "        }\n",
      "    },\n",
      "    \"prediction\": {\n",
      "        \"type\": \"neural\",\n",
      "        \"args\": {\n",
      "            \"layer_0\": {\n",
      "                \"type\": \"convolutional\",\n",
      "                \"activation\": \"relu\",\n",
      "                \"height\": 16,\n",
      "                \"width\": 16,\n",
      "                \"channels\": 1,\n",
      "                \"size\": 3,\n",
      "                \"stride\": 1,\n",
      "                \"pad\": 1,\n",
      "                \"n_init\": 24,\n",
      "                \"evolve_weights\": true,\n",
      "                \"evolve_functions\": false,\n",
      "                \"evolve_connect\": true,\n",
      "                \"evolve_neurons\": true,\n",
      "                \"n_max\": 100,\n",
      "                \"max_neuron_grow\": 1,\n",
      "                \"sgd_weights\": true,\n",
      "                \"eta\": 0.01,\n",
      "                \"evolve_eta\": true,\n",
      "                \"eta_min\": 0.0001,\n",
      "                \"momentum\": 0.9,\n",
      "                \"decay\": 0.0001\n",
      "            },\n",
      "            \"layer_1\": {\n",
      "                \"type\": \"maxpool\",\n",
      "                \"height\": 16,\n",
      "                \"width\": 16,\n",
      "                \"channels\": 24,\n",
      "                \"size\": 2,\n",
      "                \"stride\": 2\n",
      "            },\n",
      "            \"layer_2\": {\n",
      "                \"type\": \"convolutional\",\n",
      "                \"activation\": \"relu\",\n",
      "                \"height\": 8,\n",
      "                \"width\": 8,\n",
      "                \"channels\": 24,\n",
      "                \"size\": 3,\n",
      "                \"stride\": 1,\n",
      "                \"pad\": 1,\n",
      "                \"n_init\": 24,\n",
      "                \"evolve_weights\": true,\n",
      "                \"evolve_functions\": false,\n",
      "                \"evolve_connect\": true,\n",
      "                \"evolve_neurons\": true,\n",
      "                \"n_max\": 100,\n",
      "                \"max_neuron_grow\": 1,\n",
      "                \"sgd_weights\": true,\n",
      "                \"eta\": 0.01,\n",
      "                \"evolve_eta\": true,\n",
      "                \"eta_min\": 0.0001,\n",
      "                \"momentum\": 0.9,\n",
      "                \"decay\": 0.0001\n",
      "            },\n",
      "            \"layer_3\": {\n",
      "                \"type\": \"maxpool\",\n",
      "                \"height\": 8,\n",
      "                \"width\": 8,\n",
      "                \"channels\": 24,\n",
      "                \"size\": 2,\n",
      "                \"stride\": 2\n",
      "            },\n",
      "            \"layer_4\": {\n",
      "                \"type\": \"dropout\",\n",
      "                \"n_inputs\": 384,\n",
      "                \"probability\": 0.5\n",
      "            },\n",
      "            \"layer_5\": {\n",
      "                \"type\": \"connected\",\n",
      "                \"activation\": \"relu\",\n",
      "                \"n_inputs\": 384,\n",
      "                \"n_init\": 64,\n",
      "                \"evolve_weights\": true,\n",
      "                \"evolve_functions\": false,\n",
      "                \"evolve_connect\": true,\n",
      "                \"evolve_neurons\": true,\n",
      "                \"n_max\": 100,\n",
      "                \"max_neuron_grow\": 1,\n",
      "                \"sgd_weights\": true,\n",
      "                \"eta\": 0.01,\n",
      "                \"evolve_eta\": true,\n",
      "                \"eta_min\": 0.0001,\n",
      "                \"momentum\": 0.9,\n",
      "                \"decay\": 0.0001\n",
      "            },\n",
      "            \"layer_6\": {\n",
      "                \"type\": \"connected\",\n",
      "                \"activation\": \"linear\",\n",
      "                \"n_inputs\": 64,\n",
      "                \"n_init\": 10,\n",
      "                \"evolve_weights\": true,\n",
      "                \"evolve_functions\": false,\n",
      "                \"evolve_connect\": true,\n",
      "                \"evolve_neurons\": false,\n",
      "                \"sgd_weights\": true,\n",
      "                \"eta\": 0.01,\n",
      "                \"evolve_eta\": true,\n",
      "                \"eta_min\": 0.0001,\n",
      "                \"momentum\": 0.9,\n",
      "                \"decay\": 0.0001\n",
      "            },\n",
      "            \"layer_7\": {\n",
      "                \"type\": \"softmax\",\n",
      "                \"n_inputs\": 10,\n",
      "                \"scale\": 1\n",
      "            }\n",
      "        }\n",
      "    }\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "ACTIVATION: str = \"relu\"\n",
    "SGD_WEIGHTS: bool = True\n",
    "EVOLVE_WEIGHTS: bool = True\n",
    "EVOLVE_CONNECT: bool = True\n",
    "EVOLVE_ETA: bool = True\n",
    "EVOLVE_NEURONS: bool = True\n",
    "ETA: float = 0.01\n",
    "ETA_MIN: float = 0.0001\n",
    "MOMENTUM: float = 0.9\n",
    "DECAY: float = 0  # 0.0001\n",
    "N_INIT: int = 24\n",
    "N_MAX: int = 100\n",
    "MAX_GROW: int = 1\n",
    "\n",
    "LAYER_CONV: dict = {\n",
    "    \"type\": \"convolutional\",\n",
    "    \"activation\": ACTIVATION,\n",
    "    \"sgd_weights\": SGD_WEIGHTS,\n",
    "    \"evolve_weights\": EVOLVE_WEIGHTS,\n",
    "    \"evolve_connect\": EVOLVE_CONNECT,\n",
    "    \"evolve_eta\": EVOLVE_ETA,\n",
    "    \"evolve_neurons\": EVOLVE_NEURONS,\n",
    "    \"max_neuron_grow\": MAX_GROW,\n",
    "    \"eta\": ETA,\n",
    "    \"eta_min\": ETA_MIN,\n",
    "    \"momentum\": MOMENTUM,\n",
    "    \"decay\": DECAY,\n",
    "    \"n_init\": N_INIT,\n",
    "    \"n_max\": N_MAX,\n",
    "    \"stride\": 1,\n",
    "    \"size\": 3,\n",
    "    \"pad\": 1,\n",
    "    \"height\": INPUT_HEIGHT,\n",
    "    \"width\": INPUT_WIDTH,\n",
    "    \"channels\": INPUT_CHANNELS,\n",
    "}\n",
    "\n",
    "LAYER_MAXPOOL: dict = {\n",
    "    \"type\": \"maxpool\",\n",
    "    \"stride\": 2,\n",
    "    \"size\": 2,\n",
    "    \"pad\": 0,\n",
    "    \"height\": INPUT_HEIGHT,\n",
    "    \"width\": INPUT_WIDTH,\n",
    "    \"channels\": INPUT_CHANNELS,\n",
    "}\n",
    "\n",
    "LAYER_CONNECTED: dict = {\n",
    "    \"type\": \"connected\",\n",
    "    \"activation\": ACTIVATION,\n",
    "    \"sgd_weights\": SGD_WEIGHTS,\n",
    "    \"evolve_weights\": EVOLVE_WEIGHTS,\n",
    "    \"evolve_connect\": EVOLVE_CONNECT,\n",
    "    \"evolve_eta\": EVOLVE_ETA,\n",
    "    \"evolve_neurons\": EVOLVE_NEURONS,\n",
    "    \"max_neuron_grow\": MAX_GROW,\n",
    "    \"eta\": ETA,\n",
    "    \"eta_min\": ETA_MIN,\n",
    "    \"momentum\": MOMENTUM,\n",
    "    \"decay\": DECAY,\n",
    "    \"n_init\": 64,\n",
    "    \"n_max\": N_MAX,\n",
    "}\n",
    "\n",
    "E0: float = 0.01  # target error\n",
    "\n",
    "xcs = xcsf.XCS(\n",
    "    x_dim=X_DIM,\n",
    "    y_dim=Y_DIM,\n",
    "    n_actions=1,\n",
    "    omp_num_threads=12,\n",
    "    random_state=RANDOM_STATE,\n",
    "    pop_size=100,\n",
    "    max_trials=100000,\n",
    "    perf_trials=1000,\n",
    "    loss_func=\"log\",  # log loss, aka cross-entropy\n",
    "    e0=E0,\n",
    "    alpha=1,\n",
    "    beta=0.05,\n",
    "    delta=0.1,\n",
    "    theta_del=200,\n",
    "    nu=5,\n",
    "    ea={\n",
    "        \"select_type\": \"roulette\",\n",
    "        \"theta_ea\": 200,\n",
    "        \"lambda\": 2,\n",
    "        \"p_crossover\": 0.8,\n",
    "        \"err_reduc\": 1,\n",
    "        \"fit_reduc\": 0.1,\n",
    "        \"subsumption\": False,\n",
    "        \"pred_reset\": False,\n",
    "    },\n",
    "    action={\n",
    "        \"type\": \"integer\",\n",
    "    },\n",
    "    condition={\n",
    "        \"type\": \"neural\",\n",
    "        \"args\": {\n",
    "            \"layer_0\": {  # hidden layer\n",
    "                \"type\": \"connected\",\n",
    "                \"activation\": ACTIVATION,\n",
    "                \"evolve_weights\": True,\n",
    "                \"evolve_connect\": True,\n",
    "                \"evolve_neurons\": True,\n",
    "                \"n_init\": 1,\n",
    "                \"n_max\": N_MAX,\n",
    "                \"max_neuron_grow\": MAX_GROW,\n",
    "            },\n",
    "            \"layer_1\": {  # output layer\n",
    "                \"type\": \"connected\",\n",
    "                \"activation\": \"linear\",\n",
    "                \"evolve_weights\": True,\n",
    "                \"evolve_connect\": True,\n",
    "                \"n_init\": 1,\n",
    "            },\n",
    "        },\n",
    "    },\n",
    "    prediction={\n",
    "        \"type\": \"neural\",\n",
    "        \"args\": {\n",
    "            \"layer_0\": LAYER_CONV,\n",
    "            \"layer_1\": LAYER_MAXPOOL,\n",
    "            \"layer_2\": LAYER_CONV,\n",
    "            \"layer_3\": LAYER_MAXPOOL,\n",
    "            \"layer_D\": {\n",
    "                \"type\": \"dropout\",\n",
    "                \"probability\": 0.5,\n",
    "            },\n",
    "            \"layer_4\": LAYER_CONNECTED,\n",
    "            \"layer_out1\": {  # output layer - softmax composed of two layers\n",
    "                \"type\": \"connected\",\n",
    "                \"activation\": \"linear\",\n",
    "                \"sgd_weights\": SGD_WEIGHTS,\n",
    "                \"evolve_weights\": EVOLVE_WEIGHTS,\n",
    "                \"evolve_connect\": EVOLVE_CONNECT,\n",
    "                \"evolve_eta\": EVOLVE_ETA,\n",
    "                \"eta\": ETA,\n",
    "                \"eta_min\": ETA_MIN,\n",
    "                \"momentum\": MOMENTUM,\n",
    "                \"decay\": DECAY,\n",
    "                \"n_init\": Y_DIM,\n",
    "            },\n",
    "            \"layer_out2\": {  # output layer - softmax composed of two layers\n",
    "                \"type\": \"softmax\",\n",
    "                \"scale\": 1,\n",
    "            },\n",
    "        },\n",
    "    },\n",
    ")\n",
    "\n",
    "print(json.dumps(xcs.internal_params(), indent=4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Fit XCSF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2023-07-20T21:20:23.246 trials=1000 train=1.44741 val=0.82214 pset=100.0 mset=10.8 mfrac=0.76\n",
      "2023-07-20T21:20:23.269 EarlyStoppingCallback: 0.82214 best error at 1000 trials\n",
      "2023-07-20T21:20:25.713 trials=2000 train=0.70551 val=0.42768 pset=100.0 mset=22.8 mfrac=1.00\n",
      "2023-07-20T21:20:25.729 EarlyStoppingCallback: 0.42768 best error at 2000 trials\n",
      "2023-07-20T21:20:29.424 trials=3000 train=0.47797 val=0.36898 pset=100.0 mset=30.7 mfrac=1.00\n",
      "2023-07-20T21:20:29.433 EarlyStoppingCallback: 0.36898 best error at 3000 trials\n",
      "2023-07-20T21:20:34.040 trials=4000 train=0.40806 val=0.33229 pset=100.0 mset=39.6 mfrac=0.73\n",
      "2023-07-20T21:20:34.051 EarlyStoppingCallback: 0.33229 best error at 4000 trials\n",
      "2023-07-20T21:20:39.697 trials=5000 train=0.39189 val=0.31259 pset=100.0 mset=46.0 mfrac=1.00\n",
      "2023-07-20T21:20:39.708 EarlyStoppingCallback: 0.31259 best error at 5000 trials\n",
      "2023-07-20T21:20:48.540 trials=6000 train=0.33376 val=0.27304 pset=100.0 mset=51.6 mfrac=0.78\n",
      "2023-07-20T21:20:48.554 EarlyStoppingCallback: 0.27304 best error at 6000 trials\n",
      "2023-07-20T21:20:58.839 trials=7000 train=0.30711 val=0.29182 pset=100.0 mset=63.1 mfrac=1.00\n",
      "2023-07-20T21:21:12.465 trials=8000 train=0.30337 val=0.24987 pset=100.0 mset=71.3 mfrac=1.00\n",
      "2023-07-20T21:21:12.474 EarlyStoppingCallback: 0.24987 best error at 8000 trials\n",
      "2023-07-20T21:21:26.991 trials=9000 train=0.28754 val=0.23911 pset=100.0 mset=80.1 mfrac=0.74\n",
      "2023-07-20T21:21:27.004 EarlyStoppingCallback: 0.23911 best error at 9000 trials\n",
      "2023-07-20T21:21:40.655 trials=10000 train=0.26792 val=0.21554 pset=100.0 mset=88.9 mfrac=0.78\n",
      "2023-07-20T21:21:40.668 EarlyStoppingCallback: 0.21554 best error at 10000 trials\n",
      "2023-07-20T21:21:54.275 trials=11000 train=0.24129 val=0.22202 pset=100.0 mset=88.7 mfrac=1.00\n",
      "2023-07-20T21:22:07.658 trials=12000 train=0.21548 val=0.18232 pset=100.0 mset=88.6 mfrac=0.73\n",
      "2023-07-20T21:22:07.669 EarlyStoppingCallback: 0.18232 best error at 12000 trials\n",
      "2023-07-20T21:22:21.275 trials=13000 train=0.24088 val=0.21763 pset=100.0 mset=89.0 mfrac=0.84\n",
      "2023-07-20T21:22:35.066 trials=14000 train=0.20266 val=0.18681 pset=100.0 mset=89.7 mfrac=0.76\n",
      "2023-07-20T21:22:48.613 trials=15000 train=0.18357 val=0.15038 pset=100.0 mset=89.1 mfrac=0.77\n",
      "2023-07-20T21:22:48.623 EarlyStoppingCallback: 0.15038 best error at 15000 trials\n",
      "2023-07-20T21:23:02.245 trials=16000 train=0.18683 val=0.15610 pset=100.0 mset=90.8 mfrac=1.00\n",
      "2023-07-20T21:23:15.889 trials=17000 train=0.22700 val=0.16474 pset=100.0 mset=91.4 mfrac=0.75\n",
      "2023-07-20T21:23:30.003 trials=18000 train=0.19219 val=0.15873 pset=100.0 mset=90.9 mfrac=0.81\n",
      "2023-07-20T21:23:44.570 trials=19000 train=0.18603 val=0.15861 pset=100.0 mset=90.0 mfrac=0.77\n",
      "2023-07-20T21:23:59.731 trials=20000 train=0.15793 val=0.16368 pset=100.0 mset=89.8 mfrac=1.00\n",
      "2023-07-20T21:24:16.733 trials=21000 train=0.18458 val=0.13188 pset=100.0 mset=90.0 mfrac=0.84\n",
      "2023-07-20T21:24:16.744 EarlyStoppingCallback: 0.13188 best error at 21000 trials\n",
      "2023-07-20T21:24:36.632 trials=22000 train=0.14681 val=0.13445 pset=100.0 mset=89.9 mfrac=0.79\n",
      "2023-07-20T21:24:56.620 trials=23000 train=0.15355 val=0.12872 pset=100.0 mset=89.6 mfrac=0.75\n",
      "2023-07-20T21:24:56.643 EarlyStoppingCallback: 0.12872 best error at 23000 trials\n",
      "2023-07-20T21:25:16.489 trials=24000 train=0.13800 val=0.12744 pset=100.0 mset=88.9 mfrac=0.75\n",
      "2023-07-20T21:25:16.501 EarlyStoppingCallback: 0.12744 best error at 24000 trials\n",
      "2023-07-20T21:25:31.917 trials=25000 train=0.17179 val=0.13739 pset=100.0 mset=89.4 mfrac=0.74\n",
      "2023-07-20T21:25:45.684 trials=26000 train=0.14030 val=0.13156 pset=100.0 mset=89.2 mfrac=0.75\n",
      "2023-07-20T21:25:59.452 trials=27000 train=0.14652 val=0.12957 pset=100.0 mset=89.7 mfrac=0.79\n",
      "2023-07-20T21:26:13.225 trials=28000 train=0.15193 val=0.14116 pset=100.0 mset=91.0 mfrac=1.00\n",
      "2023-07-20T21:26:27.150 trials=29000 train=0.12967 val=0.12427 pset=100.0 mset=90.4 mfrac=0.75\n",
      "2023-07-20T21:26:27.161 EarlyStoppingCallback: 0.12427 best error at 29000 trials\n",
      "2023-07-20T21:26:42.044 trials=30000 train=0.13398 val=0.11724 pset=100.0 mset=90.0 mfrac=0.75\n",
      "2023-07-20T21:26:42.061 EarlyStoppingCallback: 0.11724 best error at 30000 trials\n",
      "2023-07-20T21:26:57.715 trials=31000 train=0.13790 val=0.13428 pset=100.0 mset=89.7 mfrac=0.75\n",
      "2023-07-20T21:27:15.639 trials=32000 train=0.10554 val=0.10393 pset=100.0 mset=89.9 mfrac=0.75\n",
      "2023-07-20T21:27:15.652 EarlyStoppingCallback: 0.10393 best error at 32000 trials\n",
      "2023-07-20T21:27:35.551 trials=33000 train=0.11459 val=0.11169 pset=100.0 mset=90.4 mfrac=0.75\n",
      "2023-07-20T21:27:56.393 trials=34000 train=0.11287 val=0.11577 pset=100.0 mset=91.1 mfrac=1.00\n",
      "2023-07-20T21:28:16.611 trials=35000 train=0.13269 val=0.09858 pset=100.0 mset=91.0 mfrac=1.00\n",
      "2023-07-20T21:28:16.625 EarlyStoppingCallback: 0.09858 best error at 35000 trials\n",
      "2023-07-20T21:28:32.872 trials=36000 train=0.10206 val=0.11012 pset=100.0 mset=91.0 mfrac=0.78\n",
      "2023-07-20T21:28:47.394 trials=37000 train=0.12441 val=0.10274 pset=100.0 mset=90.5 mfrac=0.67\n",
      "2023-07-20T21:29:02.706 trials=38000 train=0.12630 val=0.10549 pset=100.0 mset=90.5 mfrac=1.00\n",
      "2023-07-20T21:29:17.342 trials=39000 train=0.09984 val=0.10359 pset=100.0 mset=89.7 mfrac=0.78\n",
      "2023-07-20T21:29:32.886 trials=40000 train=0.10018 val=0.09621 pset=100.0 mset=90.3 mfrac=1.00\n",
      "2023-07-20T21:29:32.900 EarlyStoppingCallback: 0.09621 best error at 40000 trials\n",
      "2023-07-20T21:29:49.449 trials=41000 train=0.11699 val=0.10411 pset=100.0 mset=89.4 mfrac=0.75\n",
      "2023-07-20T21:30:07.592 trials=42000 train=0.10684 val=0.09631 pset=100.0 mset=88.6 mfrac=0.76\n",
      "2023-07-20T21:30:28.227 trials=43000 train=0.09634 val=0.11356 pset=100.0 mset=87.8 mfrac=0.77\n",
      "2023-07-20T21:30:49.162 trials=44000 train=0.09938 val=0.09843 pset=100.0 mset=88.5 mfrac=0.75\n",
      "2023-07-20T21:31:10.484 trials=45000 train=0.10609 val=0.10078 pset=100.0 mset=89.3 mfrac=0.78\n",
      "2023-07-20T21:31:28.372 trials=46000 train=0.11608 val=0.09928 pset=100.0 mset=89.3 mfrac=0.71\n",
      "2023-07-20T21:31:42.581 trials=47000 train=0.10817 val=0.09479 pset=100.0 mset=89.7 mfrac=0.78\n",
      "2023-07-20T21:31:42.592 EarlyStoppingCallback: 0.09479 best error at 47000 trials\n",
      "2023-07-20T21:31:57.161 trials=48000 train=0.09373 val=0.08807 pset=100.0 mset=90.1 mfrac=0.83\n",
      "2023-07-20T21:31:57.179 EarlyStoppingCallback: 0.08807 best error at 48000 trials\n",
      "2023-07-20T21:32:12.075 trials=49000 train=0.10461 val=0.09739 pset=100.0 mset=90.2 mfrac=1.00\n",
      "2023-07-20T21:32:26.631 trials=50000 train=0.10040 val=0.09908 pset=100.0 mset=90.0 mfrac=0.73\n",
      "2023-07-20T21:32:42.203 trials=51000 train=0.12531 val=0.09611 pset=100.0 mset=89.4 mfrac=0.58\n",
      "2023-07-20T21:33:00.056 trials=52000 train=0.10665 val=0.10566 pset=100.0 mset=89.7 mfrac=1.00\n",
      "2023-07-20T21:33:20.101 trials=53000 train=0.09982 val=0.13934 pset=100.0 mset=89.6 mfrac=0.74\n",
      "2023-07-20T21:33:39.925 trials=54000 train=0.09483 val=0.08828 pset=100.0 mset=89.8 mfrac=0.75\n",
      "2023-07-20T21:33:59.568 trials=55000 train=0.09823 val=0.08803 pset=100.0 mset=89.1 mfrac=0.53\n",
      "2023-07-20T21:33:59.581 EarlyStoppingCallback: 0.08803 best error at 55000 trials\n",
      "2023-07-20T21:34:17.403 trials=56000 train=0.11367 val=0.09214 pset=100.0 mset=88.7 mfrac=0.75\n",
      "2023-07-20T21:34:30.884 trials=57000 train=0.11523 val=0.08867 pset=100.0 mset=88.5 mfrac=0.74\n",
      "2023-07-20T21:34:44.622 trials=58000 train=0.10075 val=0.08936 pset=100.0 mset=90.0 mfrac=0.82\n",
      "2023-07-20T21:34:59.095 trials=59000 train=0.10174 val=0.08201 pset=100.0 mset=90.2 mfrac=0.75\n",
      "2023-07-20T21:34:59.106 EarlyStoppingCallback: 0.08201 best error at 59000 trials\n",
      "2023-07-20T21:35:14.460 trials=60000 train=0.10083 val=0.08239 pset=100.0 mset=90.4 mfrac=0.70\n",
      "2023-07-20T21:35:31.451 trials=61000 train=0.09281 val=0.08013 pset=100.0 mset=90.5 mfrac=0.78\n",
      "2023-07-20T21:35:31.466 EarlyStoppingCallback: 0.08013 best error at 61000 trials\n",
      "2023-07-20T21:35:50.875 trials=62000 train=0.10237 val=0.08532 pset=100.0 mset=90.9 mfrac=0.82\n",
      "2023-07-20T21:36:10.825 trials=63000 train=0.09335 val=0.07871 pset=100.0 mset=90.0 mfrac=1.00\n",
      "2023-07-20T21:36:10.840 EarlyStoppingCallback: 0.07871 best error at 63000 trials\n",
      "2023-07-20T21:36:30.878 trials=64000 train=0.07268 val=0.07824 pset=100.0 mset=90.6 mfrac=0.76\n",
      "2023-07-20T21:36:30.893 EarlyStoppingCallback: 0.07824 best error at 64000 trials\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2023-07-20T21:36:51.266 trials=65000 train=0.09205 val=0.07684 pset=100.0 mset=91.1 mfrac=0.75\n",
      "2023-07-20T21:36:51.281 EarlyStoppingCallback: 0.07684 best error at 65000 trials\n",
      "2023-07-20T21:37:07.028 trials=66000 train=0.08976 val=0.08443 pset=100.0 mset=90.5 mfrac=0.76\n",
      "2023-07-20T21:37:24.876 trials=67000 train=0.07414 val=0.08910 pset=100.0 mset=91.0 mfrac=0.66\n",
      "2023-07-20T21:37:43.468 trials=68000 train=0.07871 val=0.07894 pset=100.0 mset=89.9 mfrac=0.66\n",
      "2023-07-20T21:37:59.071 trials=69000 train=0.08994 val=0.09022 pset=100.0 mset=90.3 mfrac=1.00\n",
      "2023-07-20T21:38:17.594 trials=70000 train=0.09955 val=0.08027 pset=100.0 mset=90.9 mfrac=0.82\n",
      "2023-07-20T21:38:41.148 trials=71000 train=0.08883 val=0.08048 pset=100.0 mset=90.5 mfrac=0.79\n",
      "2023-07-20T21:39:04.910 trials=72000 train=0.08157 val=0.07759 pset=100.0 mset=90.5 mfrac=0.84\n",
      "2023-07-20T21:39:30.431 trials=73000 train=0.08335 val=0.07917 pset=100.0 mset=90.5 mfrac=0.79\n",
      "2023-07-20T21:39:51.065 trials=74000 train=0.08339 val=0.09963 pset=100.0 mset=91.8 mfrac=0.76\n",
      "2023-07-20T21:40:08.642 trials=75000 train=0.07723 val=0.08200 pset=100.0 mset=92.6 mfrac=0.65\n",
      "2023-07-20T21:40:27.883 trials=76000 train=0.08089 val=0.08280 pset=100.0 mset=92.3 mfrac=1.00\n",
      "2023-07-20T21:40:44.108 trials=77000 train=0.07702 val=0.08349 pset=100.0 mset=91.2 mfrac=0.72\n",
      "2023-07-20T21:41:00.213 trials=78000 train=0.09655 val=0.08136 pset=100.0 mset=91.8 mfrac=0.76\n",
      "2023-07-20T21:41:19.056 trials=79000 train=0.09386 val=0.07595 pset=100.0 mset=91.2 mfrac=0.81\n",
      "2023-07-20T21:41:19.071 EarlyStoppingCallback: 0.07595 best error at 79000 trials\n",
      "2023-07-20T21:41:40.553 trials=80000 train=0.07183 val=0.07605 pset=100.0 mset=90.8 mfrac=0.78\n",
      "2023-07-20T21:42:01.863 trials=81000 train=0.08150 val=0.07868 pset=100.0 mset=90.8 mfrac=1.00\n",
      "2023-07-20T21:42:22.946 trials=82000 train=0.07569 val=0.07681 pset=100.0 mset=89.9 mfrac=0.73\n",
      "2023-07-20T21:42:43.759 trials=83000 train=0.08836 val=0.07145 pset=100.0 mset=89.7 mfrac=0.75\n",
      "2023-07-20T21:42:43.772 EarlyStoppingCallback: 0.07145 best error at 83000 trials\n",
      "2023-07-20T21:43:00.802 trials=84000 train=0.07447 val=0.07351 pset=100.0 mset=89.7 mfrac=0.75\n",
      "2023-07-20T21:43:14.922 trials=85000 train=0.06002 val=0.06658 pset=100.0 mset=89.5 mfrac=0.78\n",
      "2023-07-20T21:43:14.934 EarlyStoppingCallback: 0.06658 best error at 85000 trials\n",
      "2023-07-20T21:43:29.369 trials=86000 train=0.06799 val=0.07405 pset=100.0 mset=90.0 mfrac=0.74\n",
      "2023-07-20T21:43:44.150 trials=87000 train=0.07594 val=0.07374 pset=100.0 mset=90.3 mfrac=0.76\n",
      "2023-07-20T21:43:59.710 trials=88000 train=0.08948 val=0.07401 pset=100.0 mset=90.0 mfrac=0.69\n",
      "2023-07-20T21:44:16.910 trials=89000 train=0.07468 val=0.07732 pset=100.0 mset=89.5 mfrac=0.65\n",
      "2023-07-20T21:44:36.432 trials=90000 train=0.07216 val=0.07080 pset=100.0 mset=88.5 mfrac=0.74\n",
      "2023-07-20T21:44:56.181 trials=91000 train=0.07759 val=0.07169 pset=100.0 mset=89.4 mfrac=1.00\n",
      "2023-07-20T21:45:16.758 trials=92000 train=0.09645 val=0.07636 pset=100.0 mset=89.3 mfrac=0.68\n",
      "2023-07-20T21:45:36.934 trials=93000 train=0.06569 val=0.07135 pset=100.0 mset=89.6 mfrac=0.76\n",
      "2023-07-20T21:45:52.740 trials=94000 train=0.06562 val=0.06925 pset=100.0 mset=89.4 mfrac=0.77\n",
      "2023-07-20T21:46:06.729 trials=95000 train=0.07762 val=0.07255 pset=100.0 mset=88.7 mfrac=1.00\n",
      "2023-07-20T21:46:20.560 trials=96000 train=0.04986 val=0.07514 pset=100.0 mset=88.7 mfrac=0.76\n",
      "2023-07-20T21:46:34.565 trials=97000 train=0.05802 val=0.06632 pset=100.0 mset=88.1 mfrac=0.76\n",
      "2023-07-20T21:46:34.575 EarlyStoppingCallback: 0.06632 best error at 97000 trials\n",
      "2023-07-20T21:46:49.629 trials=98000 train=0.06745 val=0.06836 pset=100.0 mset=88.7 mfrac=0.79\n",
      "2023-07-20T21:47:05.707 trials=99000 train=0.08868 val=0.07317 pset=100.0 mset=87.9 mfrac=0.78\n",
      "2023-07-20T21:47:24.221 trials=100000 train=0.04923 val=0.07056 pset=100.0 mset=88.3 mfrac=0.82\n",
      "2023-07-20T21:47:24.223 EarlyStoppingCallback: restoring system from trial 97000 with error=0.06632\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<xcsf.xcsf.XCS at 0x7f85ef942f70>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# use optional callback\n",
    "callback = xcsf.EarlyStoppingCallback(\n",
    "    # note: perf_trials is considered an \"epoch\" for callbacks\n",
    "    monitor=\"val\",  # which metric to monitor: {\"train\", \"val\"}\n",
    "    patience=20000,  # trials with no improvement after which training will be stopped\n",
    "    restore_best=True,  # whether to make checkpoints and restore best population\n",
    "    min_delta=0,  # minimum change to qualify as an improvement\n",
    "    start_from=0,  # trials to wait before starting to monitor improvement\n",
    "    verbose=True,  # whether to display when checkpoints are made\n",
    ")\n",
    "\n",
    "xcs.fit(\n",
    "    X_train, y_train, validation_data=(X_val, y_val), callbacks=[callback], verbose=True\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plot XCSF learning performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "metrics: dict = xcs.get_metrics()\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.plot(metrics[\"trials\"], metrics[\"train\"], label=\"Train Error\")\n",
    "plt.plot(metrics[\"trials\"], metrics[\"val\"], label=\"Validation Error\")\n",
    "plt.grid(linestyle=\"dotted\", linewidth=1)\n",
    "plt.axhline(y=E0, xmin=0, xmax=1, linestyle=\"dashed\", color=\"k\")\n",
    "plt.title(\"XCSF Training Performance\", fontsize=14)\n",
    "plt.xlabel(\"Trials\", fontsize=12)\n",
    "plt.ylabel(\"Error\", fontsize=12)\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Final XCSF test score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0     0.9882    0.9941    0.9912       169\n",
      "           1     1.0000    0.9913    0.9956       115\n",
      "           2     0.9592    0.9895    0.9741        95\n",
      "           3     0.9851    0.9167    0.9496        72\n",
      "           4     1.0000    1.0000    1.0000        84\n",
      "           5     0.9630    0.9873    0.9750        79\n",
      "           6     0.9868    0.9868    0.9868        76\n",
      "           7     0.9778    0.9888    0.9832        89\n",
      "           8     0.9706    0.9851    0.9778        67\n",
      "           9     0.9878    0.9643    0.9759        84\n",
      "\n",
      "    accuracy                         0.9828       930\n",
      "   macro avg     0.9818    0.9804    0.9809       930\n",
      "weighted avg     0.9830    0.9828    0.9827       930\n",
      "\n"
     ]
    }
   ],
   "source": [
    "pred = xcs.predict(X_test)  # soft max predictions\n",
    "pred = np.argmax(pred, axis=1)  # select most likely class\n",
    "pred = onehot_encoder.transform(pred.reshape(-1, 1))\n",
    "inv_y_test = onehot_encoder.inverse_transform(y_test)\n",
    "inv_pred = onehot_encoder.inverse_transform(pred)\n",
    "print(classification_report(inv_y_test, inv_pred, digits=4))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm = confusion_matrix(inv_y_test, inv_pred)\n",
    "disp = ConfusionMatrixDisplay(confusion_matrix=cm)\n",
    "disp.plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}