Chapter8-1 Boosting.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "d8a6d9a8",
   "metadata": {},
   "source": [
    "# Boosting 提升方法\n",
    "- 1 boosting是将弱学习算法提升为强学习算法的统计学习方。在分类学习中通过反复更新**训练数据的权重分布**，构建一些弱分类器，并将其线性组合，构造强分类器。\n",
    "\n",
    "- 2 Boosting和Bagging是集成学习的两种方式 \n",
    "     + Boosting 是每个基本模型训练时采用不同权重，针对上个一个弱分类模型修改样本权重值，序列式\n",
    "     + Bagging 是每个模型由总体样本中随机抽样得到的不同数据集训练得到的，并行式\n",
    "\n",
    "- 3 AdaBoost 模型\n",
    "$$\n",
    "f(x) = \\sum_{m=1}^{M} \\alpha_{m} G_{m} (x)\n",
    "$$\n",
    "\n",
    "- 4 AdaBoost 算法\n",
    "    + 1) 输入"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3e8d033d",
   "metadata": {},
   "source": [
    "# 例8.1 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "7789b493",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "7f51ca94",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10,) (10,)\n"
     ]
    }
   ],
   "source": [
    "X = np.array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])\n",
    "y = np.array([1, 1, 1, -1, -1, -1, 1, 1, 1, -1])\n",
    "print(X.shape, y.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "e5cdc9ce",
   "metadata": {},
   "outputs": [],
   "source": [
    "# \n",
    "\n",
    "class AdaBoost():\n",
    "    \"\"\"\n",
    "    AdaBoost\n",
    "    \"\"\"\n",
    "    \n",
    "    def calc_e_Gx(self, trainDataArr, trainLabelArr, n, div, rule, D):\n",
    "        \"\"\"\n",
    "        计算分类错误率\n",
    "        \"\"\"\n",
    "        \n",
    "        e = 0\n",
    "        \n",
    "        x = trainDataArr[:, n]\n",
    "        y = trainLabelArr\n",
    "        \n",
    "        predict = []\n",
    "        \n",
    "        if rule == 'LisOne':\n",
    "            L = 1\n",
    "            H = -1\n",
    "        else:\n",
    "            L = -1\n",
    "            H = 1\n",
    "            \n",
    "        for i in range(trainDataArr.shape[0]):\n",
    "            if x[i] < div:\n",
    "                predict.append(L)\n",
    "                if y[i] != L:\n",
    "                    e += D[i]\n",
    "            elif x[i] >= div:\n",
    "                predict.append(H)\n",
    "                if y[i] != H:\n",
    "                    e += D[i]\n",
    "                    \n",
    "        return np.array(predict), e\n",
    "    \n",
    "    def createSigleBoostingTree(self, trainDataArr, trainLabelArr, D):\n",
    "        \"\"\"\n",
    "        创建单层提升树\n",
    "        \"\"\"\n",
    "        \n",
    "        m, n = np.shape(trainDataArr)\n",
    "        \n",
    "        sigleBoostTree = {}\n",
    "        \n",
    "        sigleBoostTree['e'] = 1\n",
    "        \n",
    "        for i in range(n):\n",
    "            for div in [-0.5, 0.5, 1.5]:\n",
    "                for rule in ['LisOne', 'HisOne']:\n",
    "                    Gx, e = self.calc_e_Gx(trainDataArr, trainLabelArr, i, div, rule, D)\n",
    "                    if e < sigleBoostTree['e']:\n",
    "                        \n",
    "                        sigleBoostTree['e'] = e\n",
    "                        sigleBoostTree['div'] = div\n",
    "                        sigleBoostTree['rule'] = rule\n",
    "                        sigleBoostTree['Gx'] = Gx\n",
    "                        sigleBoostTree['feature'] = i\n",
    "                        \n",
    "        return sigleBoostTree\n",
    "    \n",
    "    \n",
    "    def createBoostingTree(self, trainDataList, trainLabelList, treeNum = 50):\n",
    "        \"\"\"\n",
    "        创建提升树\n",
    "        \"\"\"\n",
    "        \n",
    "        trainDataArr = np.array(trainDataList)\n",
    "        trainLabelArr = np.array(trainLabelList)\n",
    "        \n",
    "        finallpredict = [0] * len(trainLabelArr)\n",
    "        \n",
    "        m, n = np.shape(trainDataArr)\n",
    "        \n",
    "        D = [1 / m] * m\n",
    "        \n",
    "        tree = []\n",
    "        \n",
    "        for i in range(treeNum):\n",
    "            curTree = self.createSigleBoostingTree(trainDataArr, trainLabelArr, D)\n",
    "            alpha = 1/2 * np.log((1 - curTree['e']) / curTree['e'])\n",
    "            Gx = curTree['Gx']\n",
    "\n",
    "            D = np.multiply(D, np.exp(-1 * alpha * np.multiply(trainLabelArr, Gx))) / sum(D)\n",
    "            \n",
    "            curTree['alpha'] = alpha\n",
    "            tree.append(curTree)\n",
    "            \n",
    "            finallpredict += alpha * Gx\n",
    "            \n",
    "            error = sum([1 for i in range(len(trainDataList)) if np.sign(finallpredict[i]) != trainLabelArr[i]])\n",
    "            \n",
    "            finallError = error / len(trainDataList)\n",
    "            \n",
    "            if finallError == 0:\n",
    "                return tree\n",
    "            \n",
    "            print(\"iter : %d: %d, sigle error: %.4f, finall error: %.4f\" % (i, treeNum, curTree['e'], finallError))\n",
    "        \n",
    "        return tree\n",
    "        \n",
    "    def predict(self, x, div, rule, feature):\n",
    "        \"\"\"\n",
    "        预测标签\n",
    "        \"\"\"\n",
    "        if rule == 'LisOne':\n",
    "            L = 1\n",
    "            H = -1\n",
    "        else:\n",
    "            L = -1\n",
    "            H = 1\n",
    "            \n",
    "        if x[feature] < div:\n",
    "            return L \n",
    "        else:\n",
    "            return H\n",
    "        \n",
    "    def model_test(testDataList, testLabelList, tree):\n",
    "        \"\"\"\n",
    "        测试\n",
    "        \"\"\"\n",
    "        error_count = 0\n",
    "        for i in range(len(testDataList)):\n",
    "            result = 0\n",
    "            for curTree in tree:\n",
    "                div = curTree['div']\n",
    "                rule = curTree['rule']\n",
    "                feature = curTree['feature']\n",
    "                alpha = curTree['alpha']\n",
    "                result += alpha * self.predict(testDataList[i], div, rule, feature)\n",
    "            \n",
    "            if np.sign(result) != testLabelList[i]:\n",
    "                error_count += 1\n",
    "        \n",
    "        return 1 - error_count / len(testDataList)\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "397fb272",
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "not enough values to unpack (expected 2, got 1)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "Input \u001b[0;32mIn [5]\u001b[0m, in \u001b[0;36m<cell line: 2>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m clf \u001b[38;5;241m=\u001b[39m AdaBoost()\n\u001b[0;32m----> 2\u001b[0m tree \u001b[38;5;241m=\u001b[39m \u001b[43mclf\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mcreateBoostingTree\u001b[49m\u001b[43m(\u001b[49m\u001b[43mX\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43my\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m40\u001b[39;49m\u001b[43m)\u001b[49m\n",
      "Input \u001b[0;32mIn [3]\u001b[0m, in \u001b[0;36mAdaBoost.createBoostingTree\u001b[0;34m(self, trainDataList, trainLabelList, treeNum)\u001b[0m\n\u001b[1;32m     71\u001b[0m trainLabelArr \u001b[38;5;241m=\u001b[39m np\u001b[38;5;241m.\u001b[39marray(trainLabelList)\n\u001b[1;32m     73\u001b[0m finallpredict \u001b[38;5;241m=\u001b[39m [\u001b[38;5;241m0\u001b[39m] \u001b[38;5;241m*\u001b[39m \u001b[38;5;28mlen\u001b[39m(trainLabelArr)\n\u001b[0;32m---> 75\u001b[0m m, n \u001b[38;5;241m=\u001b[39m np\u001b[38;5;241m.\u001b[39mshape(trainDataArr)\n\u001b[1;32m     77\u001b[0m D \u001b[38;5;241m=\u001b[39m [\u001b[38;5;241m1\u001b[39m \u001b[38;5;241m/\u001b[39m m] \u001b[38;5;241m*\u001b[39m m\n\u001b[1;32m     79\u001b[0m tree \u001b[38;5;241m=\u001b[39m []\n",
      "\u001b[0;31mValueError\u001b[0m: not enough values to unpack (expected 2, got 1)"
     ]
    }
   ],
   "source": [
    "clf = AdaBoost()\n",
    "tree = clf.createBoostingTree(X, y, 40)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "67e20de1",
   "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": "3.9.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}