01 MNIST

gistfile1.txt

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'3.5.4 |Anaconda custom (64-bit)| (default, Sep 19 2017, 08:15:17) [MSC v.1900 64 bit (AMD64)]'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import sys; sys.version"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\ProgramData\\Anaconda3\\envs\\py35\\lib\\site-packages\\h5py\\__init__.py:34: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.\n",
      "  from ._conv import register_converters as _register_converters\n",
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.datasets import mnist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# images는 손글씨 숫자 이미지, labvel은 이미지의 번호\n",
    "(train_images, train_labels), (test_images, test_labels) = mnist.load_data()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### MNIST 데이터 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train_images type: <class 'numpy.ndarray'> / len: 60000\n",
      "train_labels type: <class 'numpy.ndarray'> / len: 60000\n"
     ]
    }
   ],
   "source": [
    "# train: 훈련시킹 이미지들 6만개 \n",
    "print('train_images type: {} / len: {}'.format(type(train_images), len(train_images)))\n",
    "print('train_labels type: {} / len: {}'.format(type(train_labels), len(train_labels)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test_images type: <class 'numpy.ndarray'> / len: 10000\n",
      "test_labels type: <class 'numpy.ndarray'> / len: 10000\n"
     ]
    }
   ],
   "source": [
    "# train_label : 평가 테스트용 이미지들 1만개 \n",
    "print('test_images type: {} / len: {}'.format(type(test_images), len(test_images)))\n",
    "print('test_labels type: {} / len: {}'.format(type(test_labels), len(test_labels)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(60000, 28, 28) (10000, 28, 28)\n"
     ]
    }
   ],
   "source": [
    "# 차원 확인:shape\n",
    "# (60000, 28, 28) -> 28x28 배열이 6만개 있다는 의미\n",
    "print(train_images.shape, test_images.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "- image 파일은 28x28의 배열로 이루어져 있으며 요소는 01부터 255까지 명암값으로 이루어짐.\n",
    "- 이런 이미지 파일이 6만개가 있음\n",
    "\n",
    "<img src=\"https://tensorflowkorea.gitbooks.io/tensorflow-kr/content/g3doc/images/MNIST-Matrix.png\" width=\"40%\" align=\"left\">\n",
    "<img src=\"https://tensorflowkorea.gitbooks.io/tensorflow-kr/content/g3doc/images/mnist-train-xs.png\" width=\"30%\" align=\"left\">\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Matplotlib: 배열을 이미지로 출력 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMgAAADFCAYAAAARxr1AAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIoP+nAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x18f7b160>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(3,3))\n",
    "plt.imshow(train_images[0])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 신경망 모델을 keras로 만들기"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import models, layers\n",
    "\n",
    "network = models.Sequential()\n",
    "network.add(layers.Dense(512, activation='relu', input_shape=(28*28,)))\n",
    "network.add(layers.Dense(10, activation='softmax', input_shape=(512,)))\n",
    "network.compile(\n",
    "    optimizer = 'adam',\n",
    "    loss = 'categorical_crossentropy',\n",
    "    metrics = ['accuracy'],\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 데이터 정규화"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_X = train_images.reshape(60000, -1)\n",
    "train_X = train_X.astype('float')/ 255\n",
    "test_X = test_images.reshape(10000, -1)\n",
    "test_X = test_X.astype('float')/ 255"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Reshapes** : make output to a certain shape.\n",
    "    - 0~255까지 있는 명암값을 255로 나누어 0~1까지의 수로 변환"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.utils import to_categorical\n",
    "train_Y = to_categorical(train_labels)\n",
    "test_Y = to_categorical(test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0., 0., 0., 0., 0., 1., 0., 0., 0., 0.])"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_Y[0]    # 5"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**One-Hot Encoding** : A vector that 1 in only one demension, 0 in the remaing demesion.\n",
    "    - 3일 경우 [0,0,0,1,0,0,0,0,0,0,0,0] 으로 변환. \n",
    "    - 결과적으로 tran_X는 [60000,10] 실수 배열이 된다.\n",
    "<img align=\"left\" width=\"40%\" src=\"https://tensorflowkorea.gitbooks.io/tensorflow-kr/content/g3doc/images/mnist-train-ys.png\">"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 머신러닝 훈련 과정"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "60000/60000 [==============================] - 10s 172us/step - loss: 0.2695 - acc: 0.9241\n",
      "Epoch 2/5\n",
      "60000/60000 [==============================] - 9s 157us/step - loss: 0.1077 - acc: 0.9684\n",
      "Epoch 3/5\n",
      "60000/60000 [==============================] - 9s 153us/step - loss: 0.0703 - acc: 0.9795\n",
      "Epoch 4/5\n",
      "60000/60000 [==============================] - 9s 152us/step - loss: 0.0487 - acc: 0.9855\n",
      "Epoch 5/5\n",
      "60000/60000 [==============================] - 9s 151us/step - loss: 0.0360 - acc: 0.9895\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x13f87b00>"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "network.fit(train_X, train_Y, epochs=5, batch_size=128)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 테스트셋을 사용해서 평가"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 1s 96us/step\n"
     ]
    }
   ],
   "source": [
    "test_loss, test_acc = network.evaluate(test_X, test_Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy : 0.981\n"
     ]
    }
   ],
   "source": [
    "print('Accuracy : {:.3f}'.format(test_acc))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 로스 데이터 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 1s 91us/step\n"
     ]
    }
   ],
   "source": [
    "# test_pred: 1만개의 test_X의 예측한 값\n",
    "test_pred = network.predict_classes(test_X)\n",
    "import numpy as np\n",
    "# correctly_predicted : 1만개의 예측된 값을 라벨값과 비교하여 참, 거짓 여부를 판단.\n",
    "correctly_predicted = np.equal(test_pred, test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, array([7, 2, 1, ..., 4, 5, 6], dtype=int64))"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(test_pred), test_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, array([ True,  True,  True, ...,  True,  True,  True]))"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(correctly_predicted), correctly_predicted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 149,  247,  321,  340,  381,  445,  449,  582,  613,  659,  684,\n",
       "        691,  740,  813,  877,  951,  956,  965, 1014, 1032, 1039, 1044,\n",
       "       1112, 1156, 1178, 1182, 1194, 1226, 1232, 1242, 1247, 1260, 1289,\n",
       "       1319, 1326, 1328, 1393, 1395, 1464, 1500, 1522, 1530, 1549, 1553,\n",
       "       1587, 1609, 1626, 1681, 1709, 1790, 1800, 1878, 1901, 1941, 1987,\n",
       "       2004, 2024, 2053, 2070, 2098, 2109, 2118, 2130, 2135, 2182, 2189,\n",
       "       2272, 2293, 2387, 2488, 2534, 2597, 2607, 2648, 2654, 2713, 2720,\n",
       "       2758, 2810, 2877, 2896, 2939, 2953, 3005, 3030, 3060, 3073, 3117,\n",
       "       3342, 3422, 3475, 3503, 3520, 3533, 3558, 3559, 3567, 3597, 3718,\n",
       "       3727, 3751, 3776, 3780, 3796, 3808, 3811, 3818, 3838, 3853, 3893,\n",
       "       3906, 3926, 3941, 3976, 3985, 4065, 4078, 4116, 4199, 4248, 4289,\n",
       "       4294, 4369, 4497, 4536, 4578, 4601, 4761, 4807, 4823, 4860, 4876,\n",
       "       4880, 4886, 4956, 4966, 5078, 5331, 5457, 5600, 5642, 5676, 5734,\n",
       "       5887, 5936, 5937, 5955, 5973, 6011, 6023, 6045, 6059, 6166, 6173,\n",
       "       6555, 6571, 6574, 6576, 6597, 6598, 6625, 6651, 6783, 6847, 7216,\n",
       "       7434, 7732, 7915, 8062, 8094, 8311, 8325, 8522, 8527, 9009, 9015,\n",
       "       9024, 9280, 9587, 9634, 9664, 9679, 9692, 9698, 9729, 9745, 9768,\n",
       "       9770, 9792, 9811, 9839, 9850, 9944], dtype=int64)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# wong_predictions : 거짓(라벨값과 다름)으로 판단한 숫자들의 인덱스\n",
    "wong_predictions = np.where(correctly_predicted == False)[0];\n",
    "wong_predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'plt' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-1-9e2214b791fd>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mfig\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfigure\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m: name 'plt' is not defined"
     ]
    }
   ],
   "source": [
    "fig = plt.figure()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOoAAADXCAYAAAAUcM09AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n9vgK0Qpqp1wAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1ba44c88>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOoAAADXCAYAAAAUcM09AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nwBPVuQx4ojqXgf8HzzhyGpd1aeYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1e566898>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAOoAAADXCAYAAAAUcM09AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1cc42e80>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(3):\n",
    "    wong_prediction = wong_predictions[i] # wong_prediction : wong_predictions에서 추출한 하나의 인덱스 번호\n",
    "    plt.figure(figsize=(3,3))\n",
    "    # 처음에 정의한 평가용 1만개의 이미지에 잘 못 인식한 숫자의 인덱스를 적용\n",
    "    plt.imshow(test_images[wong_prediction]) \n",
    "    # plt.title('Predicted: {}'.format(test_pred[wong_prediction]), fontsize=30)\n",
    "    plt.title('Predicted: {} / Label: {}'.format(\n",
    "        test_pred[wong_prediction], test_labels[wong_prediction] ), fontsize=20)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 신경망 추가"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {},
   "outputs": [],
   "source": [
    "network_deep = models.Sequential()\n",
    "\n",
    "network_deep.add(layers.Dense(units=512, activation='relu', input_shape=(28*28,)))\n",
    "network_deep.add(layers.Dense(units=256, activation='relu'))\n",
    "network_deep.add(layers.Dense(units=256, activation='relu'))\n",
    "network_deep.add(layers.Dense(units=10, activation='softmax'))\n",
    "                 \n",
    "network_deep.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "60000/60000 [==============================] - 13s 223us/step - loss: 0.2310 - acc: 0.9322\n",
      "Epoch 2/5\n",
      "60000/60000 [==============================] - 13s 209us/step - loss: 0.0850 - acc: 0.9737\n",
      "Epoch 3/5\n",
      "60000/60000 [==============================] - 13s 211us/step - loss: 0.0546 - acc: 0.9826\n",
      "Epoch 4/5\n",
      "60000/60000 [==============================] - 13s 213us/step - loss: 0.0399 - acc: 0.9867\n",
      "Epoch 5/5\n",
      "60000/60000 [==============================] - 13s 211us/step - loss: 0.0324 - acc: 0.9895\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x1e08da58>"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "network_deep.fit(train_X, train_Y, epochs=5, batch_size=128)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 1s 120us/step\n",
      "Accuracy of network_deep : 0.977\n"
     ]
    }
   ],
   "source": [
    "_ , acc_deep = network_deep.evaluate(test_X, test_Y)\n",
    "print('Accuracy of network_deep : {:.3f}'.format(acc_deep))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 클래스화"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "class DNN(Sequential):\n",
    "    def __init__(self, input_size, output_size, *num_hidden_nodes):\n",
    "        super().__init__()\n",
    "        num_nodes = (*num_hidden_nodes, output_size)\n",
    "        print(num_nodes)\n",
    "        \n",
    "        for idx, num_node in enumerate(num_nodes):\n",
    "            activation = 'relu'\n",
    "            if idx == 0:\n",
    "                self.add(Dense(num_node, activation = activation, input_shape=(input_size,)))\n",
    "            else:\n",
    "                if idx == len(num_nodes) - 1:\n",
    "                    activation = 'softmax'\n",
    "                self.add(Dense(output_size, activation=activation))\n",
    "        self.compile(loss='categorical_crossentropy', \n",
    "                    optimizer='adam',\n",
    "                    metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(512, 10)\n"
     ]
    }
   ],
   "source": [
    "model = DNN(train_X.shape[1], train_Y.shape[1], 512)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((784,), (10,))"
      ]
     },
     "execution_count": 161,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_X[0].shape, train_Y[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "60000/60000 [==============================] - 10s 164us/step - loss: 0.2692 - acc: 0.9242\n",
      "Epoch 2/5\n",
      "60000/60000 [==============================] - 9s 151us/step - loss: 0.1088 - acc: 0.9679\n",
      "Epoch 3/5\n",
      "60000/60000 [==============================] - 9s 150us/step - loss: 0.0702 - acc: 0.9789\n",
      "Epoch 4/5\n",
      "60000/60000 [==============================] - 9s 153us/step - loss: 0.0500 - acc: 0.9852\n",
      "Epoch 5/5\n",
      "60000/60000 [==============================] - 9s 158us/step - loss: 0.0378 - acc: 0.9887\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x1e4f5d30>"
      ]
     },
     "execution_count": 163,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(train_X, train_Y, epochs=5, batch_size=128)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 1s 98us/step\n",
      "Loss:0.06835155587559566 / Acc:0.9775\n"
     ]
    }
   ],
   "source": [
    "test_loss, test_acc = model.evaluate(test_X, test_Y)\n",
    "print('Loss:{} / Acc:{}'.format(test_loss, test_acc))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10000/10000 [==============================] - 1s 92us/step\n"
     ]
    }
   ],
   "source": [
    "# precidt : ML로 예측한 값.\n",
    "precidt = model.predict_classes(test_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(225, array([ 115,  217,  247,  321,  340,  445,  495,  522,  527,  582,  619,\n",
       "         684,  691,  720,  726,  740,  813,  846,  877,  947,  951,  956,\n",
       "         965, 1014, 1039, 1055, 1112, 1156, 1173, 1182, 1194, 1226, 1232,\n",
       "        1242, 1247, 1260, 1319, 1328, 1393, 1464, 1496, 1522, 1527, 1530,\n",
       "        1543, 1549, 1581, 1609, 1717, 1754, 1790, 1813, 1878, 1901, 1941,\n",
       "        1955, 1982, 1984, 1987, 2004, 2016, 2018, 2024, 2053, 2070, 2098,\n",
       "        2109, 2118, 2130, 2135, 2182, 2272, 2293, 2387, 2414, 2447, 2488,\n",
       "        2526, 2582, 2597, 2607, 2618, 2648, 2654, 2730, 2758, 2896, 2915,\n",
       "        2921, 2927, 2939, 2953, 2995, 3062, 3073, 3172, 3333, 3405, 3422,\n",
       "        3503, 3520, 3558, 3559, 3567, 3597, 3718, 3727, 3749, 3751, 3776,\n",
       "        3780, 3818, 3853, 3893, 3906, 3941, 3968, 3976, 4065, 4075, 4078,\n",
       "        4140, 4176, 4199, 4248, 4289, 4294, 4419, 4497, 4534, 4536, 4578,\n",
       "        4601, 4639, 4671, 4690, 4761, 4807, 4823, 4880, 4956, 4966, 5138,\n",
       "        5177, 5331, 5457, 5495, 5600, 5620, 5642, 5649, 5654, 5676, 5714,\n",
       "        5734, 5749, 5835, 5858, 5887, 5936, 5937, 5955, 5972, 5973, 5982,\n",
       "        5997, 6011, 6045, 6059, 6166, 6532, 6555, 6558, 6560, 6574, 6597,\n",
       "        6608, 6651, 6662, 6755, 6783, 6847, 7216, 7434, 7800, 7821, 7823,\n",
       "        7915, 7921, 7991, 8020, 8062, 8094, 8321, 8325, 8408, 8456, 8519,\n",
       "        8522, 8527, 9009, 9015, 9019, 9024, 9071, 9280, 9534, 9587, 9634,\n",
       "        9664, 9679, 9692, 9698, 9729, 9745, 9749, 9768, 9770, 9779, 9792,\n",
       "        9808, 9839, 9916, 9944, 9975], dtype=int64))"
      ]
     },
     "execution_count": 177,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import numpy as np\n",
    "comapare_precidt = np.equal(test_precidt, test_labels) # 예측값과 라벨값과 비교\n",
    "array_false = np.where(comapare_precidt == False)[0] # 비교해서 다른 것을 array_false에 추가.\n",
    "len(array_false), array_false"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 211,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQIAAAEKCAYAAAD5HFs9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQWBmOAjMDAeBmeEgMDPgvwE1A+ZkjVzv2AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x24732a90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQIAAAEKCAYAAAD5HFs9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nzHAQmBkOAjPDQWBmwP8DuwxihEoJGx8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x24774358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQIAAAEKCAYAAAD5HFs9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEJgZDgIzw0FgZsD/B8NyNub/qr6YAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x24700588>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQYAAAEKCAYAAADw9/tHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1ba6c668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQYAAAEKCAYAAADw9/tHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2471d860>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQYAAAEKCAYAAADw9/tHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n1Fgukn4gkwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x24d338d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQYAAAEKCAYAAADw9/tHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npxYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x208f3ef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQYAAAEKCAYAAADw9/tHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAGMxJREFUeJzt3Xu4XHV97/H3h1wIJqAEDJcQCEKqgD0GiCjGg3i4VCht\nwAIlUoRKCZ4DKmpbrOdU0kc9cpOLbcEGEwMtcpGbtFAlRBAKiuwgcosI4gYCIeGmXAIhl2//+K2B\nyfxm1uy9Z/bMED6v59nP3rO+6/Kdtdd8Zt32bEUEZmbVNuh2A2bWexwMZpZxMJhZxsFgZhkHg5ll\nHAxmlmlLMEiaL6mvTfM6SFJImtyO+Q2xh7+XdKOkFxr1Imla8bwflLRW0vwG84o6Xz+rM97bJJ0m\n6TFJr0p6RNLfDqH3+VXLWSvpUUnflbTFYOc1yOX2Va+DwW4TkiZImt3K713SfpJuk/R7ScskXS3p\n3UOYzzHF+hs31F6K+fRLOrOVeVTNKySdOMhpJjfY/kLSg2XTjmyt3dd9FdioTfPqBccDDwM3AX/a\nYJzpwIeBnwEbN5nfN4Erqh6/WF2UNAK4HtgS+L/A48AOwGaDbbzwK+AvScG/C/B1YGdJe0bE2iHO\nc7AGu01MAE4Bbgb6B7swSbsD1wHXAP9A+p18BVgg6b0R8cJg57keWArsWTNsI+AG4D/LJmxLMETE\nb9oxnx6ybUSslXQQjYPhHyPiXEjvlk3m1x8R2V5CleOA9wHvjojlxbCbB9NwjZerlne7pFeAfwV2\nB+6sHVnSKGBtRKxpYZnr6MI2cRjwHPCJiFgNIOkh4JekEC99IayPImIl6Y3rdZIOJ73uLymbdlgO\nJap2xf5Q0gJJL0v6laSP10ynYvdxuaQXJV0EbFJn/mMknS7pcUkrJf1S0oFV9Q9KWi3pU1XD3l6M\n/2+DfT4DeVdt8zvvp4DLq0Kh3RYV3ycDSLpZ0hWSZkn6DfAqsHVRe6+k64rfx4uSvi9py+qZFePc\nVhzyLJaUhWe9QwlJ20m6RNIzklZIukfSJ4rDh3uL0W6q7O4O8jmOAlZUQqHwu8qiBzmvpiSdKule\nSS9JWiLp4tr1VDXu30t6qhj3Yklvr6mPl/QvxeHPq5Jul/SBdvdcOAL4bUTcUTbScJ98/B5wLXAI\n8BBwqaRtquqfJe3uzQEOBV4BTq8znyuAY4D/D/wJ6V3vWklTAYp3xzOAsyVtW0zzLdLz+0xlJsXG\n2t+m5zYYs4vgekbSPEnjq3oaDewKVDauV4pj5O9KykJyiCYX35+qGjYd+N/AyaR1+ntJOwK3AWOA\no0jrfBfg3yWp6Hcj4EfAOOATwNeAc4BtKSFpAvBT4P3AXxfLnAtMIu3yHlmMegJp93fPqmn71eAc\nTpV/A7aWdLKkTSVNAs4iHVYtbDLtUEwgbY9/DJwEvAv4cXFYWG0msC9pr/ALxfjfqRQlbQjcCOwH\n/A1wMPA0cGOjoCmmq7z5Th5ow8X2dABN9hYAiIiWv4D5QF/V42OAAD5VNWwzYDXw6eLxCOBJ4Pya\neS0opp1cPN6nePyRmvFuAb5f9Xg0cA9pJc8opjmwZpq5wMODeF4HVfdSMl4fML9k3fwZsBdpw3ie\n9A4+oqhvVSzjRdLx8X7AsaTd4suH+rsg7S6OBqaS3o0fA95WjHMzKYS3rJn2X4EHgdFVw6YAa4A/\nLh7/H2AVsE3VONOL5zC/to+qx98AXga2atD3e4t57F2n9jAwdwDPfd9i/UbxtZh0WDjYdVjZfscN\ncPwRwMRimr2qhvcXv8dxVcOOBNYCOxWPjwVeA6ZUjTMS+A1wRtWwAE6sevxJ0utpu0E8r08W8/nD\nZuMO9x7DDZUfIuJZYDlQ2WOYRHpR/KBmmqtqHu9Leqe7TdLIyhfpXWBa1fxfIz3xvYDLgO9ExPXV\nM4qIYyNix5af1SBExDERcWVE3BIRZ5HeZXcjvWPCG3ttzwOHRcSCiJhLehc6TNIOQ1js7qQX70rg\nF8WwQyNiRdU4iyLiqZrp9gWuBtZWreffkjbwyrreo5h2SdVzvI30uy3zv4AfRsTSwT6ZiNgxIo4t\nG0fSLqQ91KuK5zGDtE6vb+OeV/XyDih2+X9PeoFW1scf1Iy6ICJeqnp8FenQ5v3F431JbxS/rVrn\nAD+havuuFREXRcTIiHh0EG3PBO6PiHubjdiuqxKN/K7m8Wuk3VRIZ+Ah36BqH29ejLuqzvxrT5bd\nAzxAOpF33qA67ZwfAi+RwuEa0sYLcFtEVD/HHxffdya9ewzGYlJIrgGeiPrnLpbVGbY56dDi5Dq1\nScX3LakfAs2CYTPqnPhso68CD1UHiKRbSS/YvyIdVrSFpPeTDpGvBk4lPfcgnegbUzP6OuslIl6R\n9BLpTRHSOv8g9bfvtp3AlbQZKYRmD2T84Q6GMpV3qwk1w2sfPwc8QTr2auZzwHtIx5X/KGmv6Nzl\nuQGJiCgO16N4vEJSvdSvnDAbSv8rIqLZlZJ6J/eeI23s36lTe6b4/hRpHdeq/b3VepY3XgzD4T2k\ny8uvi4jni3U7lL2uMoeQzgP8eRT76JK2azDuOuulOEczjnReBdI67yOd76m1si3dJoeSXu+XDmTk\nbgbD46SNbAbpXbTi4zXjLQS+CLwUEb9qNDOlG1m+Dvy/Yn6LgM+T7iHoGZI+RtowFlUN/g9ghqTR\nxSERpHMra3njbH0nLCQd6y+qbPB13AkcKWmbyuGEpOk0D4aFwGclbRER9fZWKs+79h13oB4lncR9\nXfEuOZkh3BfRxEbAqpp1dGSDcfeTNK7qcOLjpFCuBPdCYH/gsQZ7du0yE/h5DPAycteCISLWSDod\nOFPSM8CtpJN0O9WMuoB0FnyBpNOA+0mXNKcCYyLi74ozwReSjqfPinQPwinA1yRdVwkUSXNJJzFL\nzzNI+gjwTtKxOsABkp4GHoiIB4px3gl8pKhvCmwn6dDiuV1RjDOLdJx4I+kddzdScP2cdDNOxRnA\nXwBXSjqPtNt+GjAvIh6r6qsfuDkijinrvwWzK71Jmlf0PJF0QnR+RNwMfLd4DtdJmk16kXyVN/Yo\nGjmbdHhzq6Svk94YdgLGRsTppJOjrwBHF8ftqyp7PZIeBn7S5DzDt4FriqsXlwBjSYdErwEXV0Yq\n6ntHxOSmawMOlvRqzbA7SdvkSZLOAf4d+BDp91fPK6R1dQZpj+kM4OrKdgRcBHwauFnpLslHSIdd\newBPRcTZ9WYq6ZPAPGCHZucZJG0N/E/SG+zADPaMbYOznfOpf1ViXM14/cCZVY9F2qieJp2Vv5h0\ncm6dKwHAhqS72R4m/aKfIu0VVM6U/x3pjHf1md0RpMtjd/DGFYD5pJuNmj2fm3njzHb11+yqcfZu\nME5UjbMP6fLfs6RjyMdJl1HfXmeZ00jh+Arp+P8cUvBVj7McOH0wv4uS53dFg9p7SJeHnyt6eRj4\nF9a9CvE/gNtJu7oPkg7z1rkyU68PYDvSieHngRWkm4+OqKofCfy6+B1Xr8d+Glz1qZn/4aQX7gvF\nuroemFozzuWkd86BXJWo93VMMc7fFr/Pl0nBP4X8ykE/aY91dvE7fZkUWu+oWd7bgXOL+b1GOi9y\nFTC9apzaeVd6nDyA9XIS6XzT1gN9TSsa7jFaL5G0PelFOiUiHul2P29WxTmHr0TEhd3upZf5ryvf\nPD4EXONQGLpil3oUA7nB5y3OewxmlvEeg5llHAxmluno5crR2jDGMLaTizR7S3mVl3ktVrb816Qt\nBUNxs865pEuD34mIU8vGH8NYPqB9WlmkmZW4I9rzh6RDPpQobir6Z9Kfce4MzJS0c1u6MrOuauUc\nwx6kP2F+JNJtvJeSbm82sze5VoJhIulOrYolxbB1KH1KUJ+kvlVt/ZsQMxsurQRDvRMc2U0RETEn\nIqZFxLRRbNjC4sysU1oJhiW88Tf6kD6A5cnW2jGzXtBKMNwJTJG0ffG5hUeQPrzCzN7khny5MiJW\nK/0DjB+RLlfOi4j729aZmXVNS/cxRPpMxeubjmhmbyq+JdrMMg4GM8s4GMws42Aws4yDwcwyDgYz\nyzgYzCzjYDCzjIPBzDIOBjPLOBjMLONgMLOMg8HMMg4GM8s4GMws42Aws4yDwcwyDgYzyzgYzCzj\nYDCzjIPBzDIOBjPLOBjMLONgMLOMg8HMMg4GM8s4GMws42Aws4yDwcwyDgYzy4xsZWJJ/cCLwBpg\ndURMa0dTZtZdLQVD4aMR8Uwb5mNmPcKHEmaWaTUYArhB0iJJs+qNIGmWpD5JfatY2eLizKwTWj2U\nmB4RT0qaACyQ9KuIuKV6hIiYA8wB2ETjo8XlmVkHtLTHEBFPFt+XA1cDe7SjKTPrriEHg6Sxkjau\n/AzsD9zXrsbMrHtaOZTYArhaUmU+34uIH7alKzPrqiEHQ0Q8Aryvjb2YWY/w5UozyzgYzCzjYDCz\njIPBzDIOBjPLtOOPqN4SRrx7x4a1E677j9Jpv3DZX5bWN797bWn9yX2G74bRMU+WbwLvuuCRlua/\netnTjYtR/rxHbrlFaX3JEe8qra8Z3bg28bTbS6d9q/Meg5llHAxmlnEwmFnGwWBmGQeDmWUcDGaW\ncTCYWcb3MQzQqgkbN6ztNrr8s3AfOOaf291O5xzf2uS73nlkw9rq1SNKp713z4taWvZR/fs0rD17\nWkuzXu95j8HMMg4GM8s4GMws42Aws4yDwcwyDgYzyzgYzCzj+xgGaINbf9GwduAv/qp02r5p3yut\n/6zJf+77zBknlNY3WFU+fZmXJ6q0ftCMn5bWdx/bX1r/xfsvHmxLrxuh8vetJatfKq3f8cjkhrUd\neX4oLb1leI/BzDIOBjPLOBjMLONgMLOMg8HMMg4GM8s4GMws4/sY2mD8t8aW1vvnrSitH3XrZ0vr\nU84vv5egFZs1qd/zD+X1q87+89L6YYef17C2MlaXTrvb3M+V1re+5bXS+o43LiqtW2NN9xgkzZO0\nXNJ9VcPGS1og6aHi+6bD26aZddJADiXmAx+rGfYlYGFETAEWFo/NbD3RNBgi4hbguZrBM4ALi58v\nBA5uc19m1kVDPfm4RUQsBSi+T2g0oqRZkvok9a2iyR8FmFlPGParEhExJyKmRcS0UWw43IszszYY\najAsk7QVQPF9eftaMrNuG2owXAscXfx8NPCD9rRjZr2g6X0Mki4B9gY2l7QEOAU4Fbhc0rHAY8Bh\nw9lkrxu5sPx6+WW/3720PnX7x0vrK0aNLq3HqvLr+cPpC3903ZCn/dO/+HRpfbubh+/+DSvXNBgi\nYmaDUuP/5mFmb2q+JdrMMg4GM8s4GMws42Aws4yDwcwy/rPrHvD9HX5UWv+TsR8tra/53fBdrvzt\nN/YsrR+9ydml9Z1uOb5hbfufNP5Ifusu7zGYWcbBYGYZB4OZZRwMZpZxMJhZxsFgZhkHg5llfB9D\nB8y950Ol9ZM/uri0/uApO5XWd/z8zwbd0+s2GFFa3n/fu0rrG6n8T8I3+fHbGhcjSqe17vEeg5ll\nHAxmlnEwmFnGwWBmGQeDmWUcDGaWcTCYWcb3MXTApjeNKa2v/uia0vrVh5xTWv+bz39w0D1VvHpg\n+Ufbn7v1t0vrX1pWPv2Ey+5vWCt/1tZN3mMws4yDwcwyDgYzyzgYzCzjYDCzjIPBzDIOBjPL+D6G\nDthsbvm/c5++wWdL65v++tXS+gZ07/8zXPfILqX1SS/c16FOrJ2a7jFImidpuaT7qobNlvSEpLuL\nrwOHt00z66SBHErMBz5WZ/jZETG1+Lq+vW2ZWTc1DYaIuAV4rgO9mFmPaOXk44mS7ikONTZtNJKk\nWZL6JPWtYmULizOzThlqMJwP7ABMBZYC32w0YkTMiYhpETFtFBsOcXFm1klDCoaIWBYRayJiLXAB\nsEd72zKzbhpSMEjaqurhIYCvSZmtR5rexyDpEmBvYHNJS4BTgL0lTQUC6AeOH8Ye13ubXVB+n4NZ\npzUNhoiYWWfw3GHoxcx6hG+JNrOMg8HMMg4GM8s4GMws42Aws4z/7NpaokWbdLsFGwbeYzCzjIPB\nzDIOBjPLOBjMLONgMLOMg8HMMg4GM8v4PoY3gQ3GjCmtH3TXE0Oe95YjLx3ytABXHH9maf1Q/XXD\n2ugXyuc94Z9uH0pL1gbeYzCzjIPBzDIOBjPLOBjMLONgMLOMg8HMMg4GM8soIjq2sE00Pj6gfTq2\nvPWFRo0urT943vuGPO9v7HVFaf2wcc8Oed7NrKV82/v5SpXWb18xpbR+6bf2b1jbfM76+ZH9d8RC\nXojnylfcAHiPwcwyDgYzyzgYzCzjYDCzjIPBzDIOBjPLOBjMLNP0PgZJk4CLgC2BtcCciDhX0njg\nMmAy0A8cHhHPl83L9zH0nldm7FFav+m8b5fWd7r1mNL61hdt2LD2+P4jypf98fLPehi/QfnHicx6\n9ICGtWenl26qb1qdvI9hNfDFiNgJ+CBwgqSdgS8BCyNiCrCweGxm64GmwRARSyPiruLnF4HFwERg\nBnBhMdqFwMHD1aSZddagzjFImgzsCtwBbBERSyGFBzCh3c2ZWXcMOBgkjQOuBE6KiCaf1rfOdLMk\n9UnqW8XKofRoZh02oGCQNIoUChdHxFXF4GWStirqWwHL600bEXMiYlpETBtF4xNRZtY7mgaDJAFz\ngcURcVZV6Vrg6OLno4EftL89M+uGgXx8/HTgKOBeSXcXw74MnApcLulY4DHgsOFp0YaT1rQ2/YjF\n40rrG17f+CPgd7y+fN7HnfTh0vqyz3yotD5i/2ca1jZn/bxc2S5NgyEi/gtodF3UNyWYrYd856OZ\nZRwMZpZxMJhZxsFgZhkHg5llHAxmlvHHx7/FjZy0TWl9xg13ldbfMeLl0vq8mQc1rMWi+0untcHz\nx8eb2bBxMJhZxsFgZhkHg5llHAxmlnEwmFnGwWBmmYF8HoOtx1Y/vqS0fval5Z/xe9+sfyqtn/zp\ntzWs/cFxpZNaF3mPwcwyDgYzyzgYzCzjYDCzjIPBzDIOBjPLOBjMLOP7GKzU9t9+uLR+5ZGbltYv\n3OeChrWv7Xl0wxqAfvrL0roNH+8xmFnGwWBmGQeDmWUcDGaWcTCYWcbBYGYZB4OZZZrexyBpEnAR\nsCWwFpgTEedKmg0cBzxdjPrliLh+uBq17lizbHlp/bQH/6i0/vPdLm1Ye/ErL5VOu8kBpWUbRgO5\nwWk18MWIuEvSxsAiSQuK2tkRcebwtWdm3dA0GCJiKbC0+PlFSYuBicPdmJl1z6DOMUiaDOwK3FEM\nOlHSPZLmSap7b6ykWZL6JPWtYmVLzZpZZww4GCSNA64EToqIF4DzgR2AqaQ9im/Wmy4i5kTEtIiY\nNooN29CymQ23AQWDpFGkULg4Iq4CiIhlEbEmItYCFwB7DF+bZtZJTYNBkoC5wOKIOKtq+FZVox0C\n3Nf+9sysGwZyVWI6cBRwr6S7i2FfBmZKmgoE0A8cPywdWk97x9kbl9aXzl/RsHbZLheWTnscHx5S\nT9a6gVyV+C9AdUq+Z8FsPeU7H80s42Aws4yDwcwyDgYzyzgYzCzjYDCzjD8+3loy8seLSuvHbut7\nEd6MvMdgZhkHg5llHAxmlnEwmFnGwWBmGQeDmWUcDGaWUUR0bmHS08CjVYM2B57pWAOD06u99Wpf\n4N6Gqp29bRcR72x1Jh0NhmzhUl9ETOtaAyV6tbde7Qvc21D1Ym8+lDCzjIPBzDLdDoY5XV5+mV7t\nrVf7Avc2VD3XW1fPMZhZb+r2HoOZ9SAHg5lluhIMkj4m6UFJD0v6Ujd6aERSv6R7Jd0tqa/LvcyT\ntFzSfVXDxktaIOmh4nvd/xnapd5mS3qiWHd3SzqwS71NknSTpMWS7pf0uWJ4V9ddSV89sd7W6bXT\n5xgkjQB+DewHLAHuBGZGxAMdbaQBSf3AtIjo+s0wkvYCXgIuioj3FsNOB56LiFOLUN00Ik7ukd5m\nAy9FxJmd7qemt62ArSLiLkkbA4uAg4Fj6OK6K+nrcHpgvVXrxh7DHsDDEfFIRLwGXArM6EIfPS8i\nbgGeqxk8A6j8C6cLSRtWxzXorSdExNKIuKv4+UVgMTCRLq+7kr56TjeCYSLweNXjJfTWygngBkmL\nJM3qdjN1bBERSyFtaMCELvdT60RJ9xSHGl05zKkmaTKwK3AHPbTuavqCHltv3QiGev/urpeumU6P\niN2AA4ATil1mG5jzgR2AqcBS4JvdbEbSONJ/aT8pIl7oZi/V6vTVU+sNuhMMS4BJVY+3AZ7sQh91\nRcSTxfflwNWkQ59esqzyn8aL78u73M/rImJZRKyJiLXABXRx3UkaRXrxXRwRVxWDu77u6vXVS+ut\nohvBcCcwRdL2kkYDRwDXdqGPjKSxxUkhJI0F9gfuK5+q464Fji5+Phr4QRd7WUflRVc4hC6tO0kC\n5gKLI+KsqlJX112jvnplvVXryp2PxeWYc4ARwLyI+HrHm6hD0rtIewmQPlr/e93sTdIlwN6kP8td\nBpwCXANcDmwLPAYcFhEdPwnYoLe9SbvDAfQDx1eO6Tvc24eBW4F7gbXF4C+Tjue7tu5K+ppJD6y3\nar4l2swyvvPRzDIOBjPLOBjMLONgMLOMg8HMMg4GM8s4GMws89/D1WRdgxKWKQAAAABJRU5ErkJg\ngg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x20894a20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQYAAAEKCAYAAADw9/tHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nwWBmGQeDmWUcDGaWcTCYWeb/A0z9UEhln29gAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x2050b978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQYAAAEKCAYAAADw9/tHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x246f8d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 잘못 인식한 숫자중 하나를 추출해서 이미지화\n",
    "for i in range(20,30):\n",
    "    index = array_false[i]\n",
    "    select = precidt[index]\n",
    "    label = test_labels[index]\n",
    "\n",
    "    plt.imshow(test_images[index])\n",
    "    plt.title('index: {}, Predict: {}, Label: {}'.format(index, select, label), fontsize=15)\n",
    "    plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "python3.5",
   "language": "python",
   "name": "python3.5"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}