11장 다층 퍼셉트론 모델.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [예제 11.1] MLPClassifier의 최적화 알고리즘 비교하기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn import datasets\n",
    "from sklearn.preprocessing import MinMaxScaler\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from sklearn.exceptions import ConvergenceWarning\n",
    "%matplotlib inline\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "iris = datasets.load_iris()\n",
    "X_digits, y_digits = datasets.load_digits(return_X_y=True)\n",
    "data_sets = [(iris.data, iris.target), (X_digits, y_digits),\n",
    "             datasets.make_circles(noise=0.2, factor=0.5, random_state=1),\n",
    "             datasets.make_moons(noise=0.3, random_state=0)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "labels = [\"constant learning-rate\", \"inv-scaling learning-rate\", \"adam\"]\n",
    "params = [{\n",
    "    \"solver\": \"sgd\",\n",
    "    \"learning_rate\": \"constant\",\n",
    "    \"momentum\": 0,\n",
    "    \"learning_rate_init\": 0.2\n",
    "}, {\n",
    "    \"solver\": \"sgd\",\n",
    "    \"learning_rate\": \"invscaling\",\n",
    "    \"momentum\": 0,\n",
    "    \"learning_rate_init\": 0.2\n",
    "}, {\n",
    "    \"solver\": \"adam\",\n",
    "    \"learning_rate_init\": 0.01\n",
    "}]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_args = [{\n",
    "    \"c\": \"red\",\n",
    "    \"linestyle\": \"-\"\n",
    "}, {\n",
    "    \"c\": \"blue\",\n",
    "    \"linestyle\": \"-\"\n",
    "}, {\n",
    "    \"c\": \"black\",\n",
    "    \"linestyle\": \"-\"\n",
    "}]\n",
    "\n",
    "\n",
    "def plot_on_dataset(X, y, ax, name):\n",
    "    ax.set_title(name)\n",
    "\n",
    "    X = MinMaxScaler().fit_transform(X)\n",
    "    mlps = []\n",
    "    if name == \"digits\": max_iter = 15\n",
    "    else: max_iter = 400\n",
    "\n",
    "    for label, param in zip(labels, params):\n",
    "        mlp = MLPClassifier(random_state=0, max_iter=max_iter,\n",
    "                            **param).fit(X, y)\n",
    "        mlps.append(mlp)\n",
    "\n",
    "    for mlp, label, args in zip(mlps, labels, plot_args):\n",
    "        ax.plot(mlp.loss_curve_, label=label, **args)\n",
    "\n",
    "\n",
    "fig, axes = plt.subplots(2, 2, figsize=(15, 10))\n",
    "for ax, data, name in zip(axes.ravel(), data_sets,\n",
    "                          [\"iris\", \"digits\", \"circles\", \"moons\"]):\n",
    "    plot_on_dataset(*data, ax=ax, name=name)\n",
    "\n",
    "fig.legend(ax.get_lines(), labels, ncol=3, loc=\"upper center\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 11 되새김 문제"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [되새김 문제 11.1] MLPClassifier 클래스 적용하기"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 문제"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import load_breast_cancer\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "\n",
    "X, y = load_breast_cancer(return_X_y=True, as_frame=False)\n",
    "X_train, X_test, y_train, y_test = train_test_split(X,\n",
    "                                                    y,\n",
    "                                                    test_size=0.30,\n",
    "                                                    random_state=1234)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 풀이"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "테스트셋 재현율: 91.43%\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import recall_score\n",
    "\n",
    "clf = MLPClassifier(hidden_layer_sizes=(20, 10),\n",
    "                    solver='adam',\n",
    "                    max_iter=1000,\n",
    "                    random_state=1234,\n",
    "                    early_stopping=True)\n",
    "\n",
    "y_pred = clf.fit(X_train, y_train).predict(X_test)\n",
    "\n",
    "print(f'테스트셋 재현율:{recall_score(y_test, y_pred) * 100: .2f}%')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### [되새김 문제 11.2] 최적의 지도 학습 모델 탐색하기"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 문제"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import load_digits\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "\n",
    "X, y = load_digits(return_X_y=True)\n",
    "X_train, X_test, y_train, y_test = train_test_split(X,\n",
    "                                                    y,\n",
    "                                                    test_size=0.25,\n",
    "                                                    random_state=1234)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 풀이"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "모델: logistic, 소요 시간: 2.386초, 탐색 결과: {'logistic__C': 0.046415888336127774, 'logistic__penalty': 'l2'}, CV 점수: 0.923\n",
      "모델: DT, 소요 시간: 4.749초, 탐색 결과: {'DT__ccp_alpha': 0.0026826957952797246, 'DT__max_depth': 5}, CV 점수: 0.629\n",
      "모델: RF, 소요 시간: 23.586초, 탐색 결과: {'RF__max_depth': 11}, CV 점수: 0.939\n",
      "모델: KNN, 소요 시간: 2.469초, 탐색 결과: {'KNN__n_neighbors': 3}, CV 점수: 0.945\n",
      "모델: svc, 소요 시간: 25.631초, 탐색 결과: {'svc__C': 21.54434690031882, 'svc__kernel': 'poly'}, CV 점수: 0.952\n",
      "모델: mlp, 소요 시간: 95.920초, 탐색 결과: {'mlp__hidden_layer_sizes': (30,), 'mlp__solver': 'adam'}, CV 점수: 0.927\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "from sklearn.pipeline import Pipeline\n",
    "import time\n",
    "import warnings\n",
    "\n",
    "warnings.filterwarnings(\"ignore\")\n",
    "\n",
    "scaler = StandardScaler()\n",
    "\n",
    "logistic = LogisticRegression(max_iter=10000, tol=0.1, random_state=1234)\n",
    "DT = DecisionTreeClassifier(random_state=1234)\n",
    "RF = RandomForestClassifier(random_state=1234, n_estimators=200)\n",
    "KNN = KNeighborsClassifier()\n",
    "svc = SVC(random_state=1234)\n",
    "mlp = MLPClassifier(random_state=1234)\n",
    "\n",
    "logistic_param_grid = {\n",
    "    'logistic__penalty': ['l1', 'l2', 'elasticnet'],\n",
    "    'logistic__C': np.logspace(-4, 4, 4)\n",
    "}\n",
    "DT_param_grid = {\n",
    "    'DT__max_depth': range(1, 6),\n",
    "    'DT__ccp_alpha': np.logspace(-6, 2, 8)\n",
    "}\n",
    "RF_param_grid = {'RF__max_depth': [9, 10, 11]}\n",
    "KNN_param_grid = {'KNN__n_neighbors': [2 * x + 1 for x in range(1, 10)]}\n",
    "svc_param_grid = {\n",
    "    'svc__C': np.logspace(-4, 4, 4),\n",
    "    'svc__kernel': ['linear', 'poly', 'rbf', 'sigmoid']\n",
    "}\n",
    "mlp_param_grid = {\n",
    "    'mlp__hidden_layer_sizes': [(30, ), (20, 20), (10, 10, 10)],\n",
    "    'mlp__solver': ['lbfgs', 'sgd', 'adam']\n",
    "}\n",
    "\n",
    "search_space = [['logistic', logistic, logistic_param_grid],\n",
    "                ['DT', DT, DT_param_grid], ['RF', RF, RF_param_grid],\n",
    "                ['KNN', KNN, KNN_param_grid], ['svc', svc, svc_param_grid],\n",
    "                ['mlp', mlp, mlp_param_grid]]\n",
    "\n",
    "for s in search_space:\n",
    "    start_time = time.time()\n",
    "    pipe = Pipeline(steps=[(\"scaler\", scaler), (s[0], s[1])])\n",
    "    search = GridSearchCV(pipe, s[2])\n",
    "    search.fit(X, y)\n",
    "\n",
    "    print(f'모델: {s[0]}, 소요 시간: {time.time() - start_time:.3f}초, 탐색 결과: {search.best_params_}, CV 점수: {search.best_score_:.3f}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "테스트셋 정확도: 98.89%\n"
     ]
    }
   ],
   "source": [
    "clf = SVC(C=21.54434690031882, kernel='poly',\n",
    "          random_state=1234).fit(X_train, y_train)\n",
    "\n",
    "y_pred = clf.predict(X_test)\n",
    "print(f'테스트셋 정확도:{(y_pred == y_test).mean() * 100: .2f}%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}