chapter10_dl-for-timeseries.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rLyx0NKfjbcR"
      },
      "source": [
        "이 노트북은 [케라스 창시자에게 배우는 딥러닝 2판](https://tensorflow.blog/kerasdl2/)의 예제 코드를 담고 있습니다.\n",
        "\n",
        "<table align=\"left\">\n",
        "    <tr>\n",
        "        <td>\n",
        "            <a href=\"https://colab.research.google.com/github/rickiepark/deep-learning-with-python-2nd/blob/main/chapter10_dl-for-timeseries.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
        "        </td>\n",
        "    </tr>\n",
        "</table>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-jmgSYTnjbcU"
      },
      "source": [
        "# 시계열을 위한 딥러닝"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yqfO6TQzjbcV"
      },
      "source": [
        "## 다양한 종류의 시계열 작업"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rMlIvey2jbcV"
      },
      "source": [
        "## 기온 예측 문제"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:26:55.453195Z",
          "iopub.status.busy": "2022-05-20T13:26:55.452690Z",
          "iopub.status.idle": "2022-05-20T13:26:58.879200Z",
          "shell.execute_reply": "2022-05-20T13:26:58.877667Z"
        },
        "id": "aUCXJRQgjbcV",
        "outputId": "931a57f7-a0d8-4a45-f897-55d8f0d074d1",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "--2022-11-06 05:02:22--  https://s3.amazonaws.com/keras-datasets/jena_climate_2009_2016.csv.zip\n",
            "Resolving s3.amazonaws.com (s3.amazonaws.com)... 52.216.49.64\n",
            "Connecting to s3.amazonaws.com (s3.amazonaws.com)|52.216.49.64|:443... connected.\n",
            "HTTP request sent, awaiting response... 200 OK\n",
            "Length: 13565642 (13M) [application/zip]\n",
            "Saving to: ‘jena_climate_2009_2016.csv.zip’\n",
            "\n",
            "jena_climate_2009_2 100%[===================>]  12.94M  43.3MB/s    in 0.3s    \n",
            "\n",
            "2022-11-06 05:02:22 (43.3 MB/s) - ‘jena_climate_2009_2016.csv.zip’ saved [13565642/13565642]\n",
            "\n",
            "Archive:  jena_climate_2009_2016.csv.zip\n",
            "  inflating: jena_climate_2009_2016.csv  \n",
            "  inflating: __MACOSX/._jena_climate_2009_2016.csv  \n"
          ]
        }
      ],
      "source": [
        "!wget https://s3.amazonaws.com/keras-datasets/jena_climate_2009_2016.csv.zip\n",
        "!unzip jena_climate_2009_2016.csv.zip"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "C93ZoFD3jbcX"
      },
      "source": [
        "**예나 날씨 데이터셋 조사하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:26:58.883391Z",
          "iopub.status.busy": "2022-05-20T13:26:58.883062Z",
          "iopub.status.idle": "2022-05-20T13:26:59.026206Z",
          "shell.execute_reply": "2022-05-20T13:26:59.025347Z"
        },
        "id": "Z-0eD31YjbcX",
        "outputId": "f3654dc6-66c6-4e79-ff73-0d3937559cd4",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "['\"Date Time\"', '\"p (mbar)\"', '\"T (degC)\"', '\"Tpot (K)\"', '\"Tdew (degC)\"', '\"rh (%)\"', '\"VPmax (mbar)\"', '\"VPact (mbar)\"', '\"VPdef (mbar)\"', '\"sh (g/kg)\"', '\"H2OC (mmol/mol)\"', '\"rho (g/m**3)\"', '\"wv (m/s)\"', '\"max. wv (m/s)\"', '\"wd (deg)\"']\n",
            "420451\n"
          ]
        }
      ],
      "source": [
        "import os\n",
        "fname = os.path.join(\"jena_climate_2009_2016.csv\")\n",
        "\n",
        "with open(fname) as f:\n",
        "    data = f.read()\n",
        "\n",
        "lines = data.split(\"\\n\")\n",
        "header = lines[0].split(\",\")\n",
        "lines = lines[1:]\n",
        "print(header)\n",
        "print(len(lines))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kRnwyaNijbcY"
      },
      "source": [
        "**데이터 파싱**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:26:59.030213Z",
          "iopub.status.busy": "2022-05-20T13:26:59.029492Z",
          "iopub.status.idle": "2022-05-20T13:27:00.936580Z",
          "shell.execute_reply": "2022-05-20T13:27:00.935499Z"
        },
        "id": "FiZFxhiCjbcY"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "temperature = np.zeros((len(lines),))\n",
        "raw_data = np.zeros((len(lines), len(header) - 1))\n",
        "for i, line in enumerate(lines):\n",
        "    values = [float(x) for x in line.split(\",\")[1:]]\n",
        "    temperature[i] = values[1]\n",
        "    raw_data[i, :] = values[:]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "b1MbsjsljbcZ"
      },
      "source": [
        "**전체 기온을 그래프로 그리기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:27:00.940623Z",
          "iopub.status.busy": "2022-05-20T13:27:00.940326Z",
          "iopub.status.idle": "2022-05-20T13:27:02.011631Z",
          "shell.execute_reply": "2022-05-20T13:27:02.010321Z"
        },
        "id": "o6gAh5ecjbcZ",
        "outputId": "1078d50c-da15-4a0b-b0a9-8d043830f62e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "from matplotlib import pyplot as plt\n",
        "plt.plot(range(len(temperature)), temperature)\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aALKlaFJjbcZ"
      },
      "source": [
        "**처음 10일간의 기온을 그래프로 그리기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:27:02.015866Z",
          "iopub.status.busy": "2022-05-20T13:27:02.015374Z",
          "iopub.status.idle": "2022-05-20T13:27:02.123996Z",
          "shell.execute_reply": "2022-05-20T13:27:02.122754Z"
        },
        "id": "FGPdWjXujbca",
        "outputId": "330946c0-e1be-4bc9-b6a6-46b72938acd2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plt.plot(range(1440), temperature[:1440])\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QaZN550Xjbca"
      },
      "source": [
        "**각 분할에 사용할 샘플 수 계산하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:27:02.128070Z",
          "iopub.status.busy": "2022-05-20T13:27:02.127766Z",
          "iopub.status.idle": "2022-05-20T13:27:02.133998Z",
          "shell.execute_reply": "2022-05-20T13:27:02.132864Z"
        },
        "id": "YG7JjzJtjbca",
        "outputId": "c4e2a62b-2b39-45eb-da62-3c058fbec173",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "num_train_samples: 210225\n",
            "num_val_samples: 105112\n",
            "num_test_samples: 105114\n"
          ]
        }
      ],
      "source": [
        "num_train_samples = int(0.5 * len(raw_data))\n",
        "num_val_samples = int(0.25 * len(raw_data))\n",
        "num_test_samples = len(raw_data) - num_train_samples - num_val_samples\n",
        "print(\"num_train_samples:\", num_train_samples)\n",
        "print(\"num_val_samples:\", num_val_samples)\n",
        "print(\"num_test_samples:\", num_test_samples)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "B-GzlCYCjbcb"
      },
      "source": [
        "### 데이터 준비"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gsmS_D67jbcb"
      },
      "source": [
        "**데이터 정규화**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:27:02.137634Z",
          "iopub.status.busy": "2022-05-20T13:27:02.137353Z",
          "iopub.status.idle": "2022-05-20T13:27:02.191157Z",
          "shell.execute_reply": "2022-05-20T13:27:02.189956Z"
        },
        "id": "5BhCfjkrjbcb"
      },
      "outputs": [],
      "source": [
        "mean = raw_data[:num_train_samples].mean(axis=0)\n",
        "raw_data -= mean\n",
        "std = raw_data[:num_train_samples].std(axis=0)\n",
        "raw_data /= std"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:27:02.195344Z",
          "iopub.status.busy": "2022-05-20T13:27:02.195011Z",
          "iopub.status.idle": "2022-05-20T13:27:02.314820Z",
          "shell.execute_reply": "2022-05-20T13:27:02.312523Z"
        },
        "id": "sgoxHRsPjbcb",
        "outputId": "25ce4f20-9b3e-4693-95ec-59ed2087e041",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0, 1, 2] 3\n",
            "[1, 2, 3] 4\n",
            "[2, 3, 4] 5\n",
            "[3, 4, 5] 6\n",
            "[4, 5, 6] 7\n"
          ]
        }
      ],
      "source": [
        "import numpy as np\n",
        "from tensorflow import keras\n",
        "int_sequence = np.arange(10)\n",
        "dummy_dataset = keras.utils.timeseries_dataset_from_array(\n",
        "    data=int_sequence[:-3],\n",
        "    targets=int_sequence[3:],\n",
        "    sequence_length=3,\n",
        "    batch_size=2,\n",
        ")\n",
        "\n",
        "for inputs, targets in dummy_dataset:\n",
        "    for i in range(inputs.shape[0]):\n",
        "        print([int(x) for x in inputs[i]], int(targets[i]))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "n_d8n6Yqjbcb"
      },
      "source": [
        "**훈련, 검증, 테스트 데이터셋 만들기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:27:02.318918Z",
          "iopub.status.busy": "2022-05-20T13:27:02.318531Z",
          "iopub.status.idle": "2022-05-20T13:27:02.483352Z",
          "shell.execute_reply": "2022-05-20T13:27:02.482263Z"
        },
        "id": "uts3EvXajbcc"
      },
      "outputs": [],
      "source": [
        "sampling_rate = 6\n",
        "sequence_length = 120\n",
        "delay = sampling_rate * (sequence_length + 24 - 1)\n",
        "batch_size = 256\n",
        "\n",
        "train_dataset = keras.utils.timeseries_dataset_from_array(\n",
        "    raw_data[:-delay],\n",
        "    targets=temperature[delay:],\n",
        "    sampling_rate=sampling_rate,\n",
        "    sequence_length=sequence_length,\n",
        "    shuffle=True,\n",
        "    batch_size=batch_size,\n",
        "    start_index=0,\n",
        "    end_index=num_train_samples)\n",
        "\n",
        "val_dataset = keras.utils.timeseries_dataset_from_array(\n",
        "    raw_data[:-delay],\n",
        "    targets=temperature[delay:],\n",
        "    sampling_rate=sampling_rate,\n",
        "    sequence_length=sequence_length,\n",
        "    shuffle=True,\n",
        "    batch_size=batch_size,\n",
        "    start_index=num_train_samples,\n",
        "    end_index=num_train_samples + num_val_samples)\n",
        "\n",
        "test_dataset = keras.utils.timeseries_dataset_from_array(\n",
        "    raw_data[:-delay],\n",
        "    targets=temperature[delay:],\n",
        "    sampling_rate=sampling_rate,\n",
        "    sequence_length=sequence_length,\n",
        "    shuffle=True,\n",
        "    batch_size=batch_size,\n",
        "    start_index=num_train_samples + num_val_samples)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dZNUb2s2jbcc"
      },
      "source": [
        "**훈련 데이터셋의 배치 크기 확인하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:27:02.487516Z",
          "iopub.status.busy": "2022-05-20T13:27:02.487209Z",
          "iopub.status.idle": "2022-05-20T13:27:02.708306Z",
          "shell.execute_reply": "2022-05-20T13:27:02.707147Z"
        },
        "id": "pC_5Co8Zjbcc",
        "outputId": "7bb06d94-0563-4a33-91fc-4894dae2aa9f",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "샘플 크기: (256, 120, 14)\n",
            "타깃 크기: (256,)\n"
          ]
        }
      ],
      "source": [
        "for samples, targets in train_dataset:\n",
        "    print(\"샘플 크기:\", samples.shape)\n",
        "    print(\"타깃 크기:\", targets.shape)\n",
        "    break"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vdYKKLDjjbcd"
      },
      "source": [
        "### 상식 수준의 기준점"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OABELWY3jbcd"
      },
      "source": [
        "**상식적인 기준 모델의 MAE 계산하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:27:02.712723Z",
          "iopub.status.busy": "2022-05-20T13:27:02.712259Z",
          "iopub.status.idle": "2022-05-20T13:27:12.831040Z",
          "shell.execute_reply": "2022-05-20T13:27:12.829972Z"
        },
        "id": "-ryWeRYljbcd",
        "outputId": "acb163c9-ff22-4ae2-ab4a-06a7380f0a71",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "검증 MAE: 2.44\n",
            "테스트 MAE: 2.62\n"
          ]
        }
      ],
      "source": [
        "def evaluate_naive_method(dataset):\n",
        "    total_abs_err = 0.\n",
        "    samples_seen = 0\n",
        "    for samples, targets in dataset:\n",
        "        preds = samples[:, -1, 1] * std[1] + mean[1]\n",
        "        total_abs_err += np.sum(np.abs(preds - targets))\n",
        "        samples_seen += samples.shape[0]\n",
        "    return total_abs_err / samples_seen\n",
        "\n",
        "print(f\"검증 MAE: {evaluate_naive_method(val_dataset):.2f}\")\n",
        "print(f\"테스트 MAE: {evaluate_naive_method(test_dataset):.2f}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AwYpH902jbcd"
      },
      "source": [
        "### 기본적인 머신 러닝 모델 시도해 보기"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "g8shk8fVjbcd"
      },
      "source": [
        "**밀집 연결 모델 훈련하고 평가하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:27:12.835455Z",
          "iopub.status.busy": "2022-05-20T13:27:12.834662Z",
          "iopub.status.idle": "2022-05-20T13:30:16.406591Z",
          "shell.execute_reply": "2022-05-20T13:30:16.405650Z"
        },
        "id": "AnkykZycjbcd",
        "outputId": "83d57b68-6f2b-414d-dbae-9b4cb959aade",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "819/819 [==============================] - 15s 14ms/step - loss: 13.6516 - mae: 2.8433 - val_loss: 12.2475 - val_mae: 2.7247\n",
            "Epoch 2/10\n",
            "819/819 [==============================] - 11s 13ms/step - loss: 9.5950 - mae: 2.4328 - val_loss: 11.7673 - val_mae: 2.7202\n",
            "Epoch 3/10\n",
            "819/819 [==============================] - 11s 13ms/step - loss: 8.7626 - mae: 2.3289 - val_loss: 11.8227 - val_mae: 2.7062\n",
            "Epoch 4/10\n",
            "819/819 [==============================] - 11s 13ms/step - loss: 8.2281 - mae: 2.2607 - val_loss: 12.3838 - val_mae: 2.8054\n",
            "Epoch 5/10\n",
            "819/819 [==============================] - 11s 13ms/step - loss: 7.8869 - mae: 2.2140 - val_loss: 10.4159 - val_mae: 2.5410\n",
            "Epoch 6/10\n",
            "819/819 [==============================] - 11s 13ms/step - loss: 7.6061 - mae: 2.1748 - val_loss: 10.7341 - val_mae: 2.5912\n",
            "Epoch 7/10\n",
            "819/819 [==============================] - 11s 13ms/step - loss: 7.4322 - mae: 2.1493 - val_loss: 10.6596 - val_mae: 2.5763\n",
            "Epoch 8/10\n",
            "819/819 [==============================] - 11s 13ms/step - loss: 7.2412 - mae: 2.1219 - val_loss: 10.9985 - val_mae: 2.6152\n",
            "Epoch 9/10\n",
            "819/819 [==============================] - 11s 13ms/step - loss: 7.0759 - mae: 2.0976 - val_loss: 10.6195 - val_mae: 2.5765\n",
            "Epoch 10/10\n",
            "819/819 [==============================] - 11s 13ms/step - loss: 6.9707 - mae: 2.0840 - val_loss: 10.6545 - val_mae: 2.5767\n",
            "405/405 [==============================] - 4s 9ms/step - loss: 11.4027 - mae: 2.6500\n",
            "테스트 MAE: 2.65\n"
          ]
        }
      ],
      "source": [
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "\n",
        "inputs = keras.Input(shape=(sequence_length, raw_data.shape[-1]))\n",
        "x = layers.Flatten()(inputs)\n",
        "x = layers.Dense(16, activation=\"relu\")(x)\n",
        "outputs = layers.Dense(1)(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "\n",
        "callbacks = [\n",
        "    keras.callbacks.ModelCheckpoint(\"jena_dense.keras\",\n",
        "                                    save_best_only=True)\n",
        "]\n",
        "model.compile(optimizer=\"rmsprop\", loss=\"mse\", metrics=[\"mae\"])\n",
        "history = model.fit(train_dataset,\n",
        "                    epochs=10,\n",
        "                    validation_data=val_dataset,\n",
        "                    callbacks=callbacks)\n",
        "\n",
        "model = keras.models.load_model(\"jena_dense.keras\")\n",
        "print(f\"테스트 MAE: {model.evaluate(test_dataset)[1]:.2f}\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lmTjKA_qjbce"
      },
      "source": [
        "**결과 그래프 그리기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:30:16.410386Z",
          "iopub.status.busy": "2022-05-20T13:30:16.410076Z",
          "iopub.status.idle": "2022-05-20T13:30:16.562302Z",
          "shell.execute_reply": "2022-05-20T13:30:16.561390Z"
        },
        "id": "GE_ngKAZjbce",
        "outputId": "476e7c37-3db7-4692-ce02-077c70f702a6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "import matplotlib.pyplot as plt\n",
        "loss = history.history[\"mae\"]\n",
        "val_loss = history.history[\"val_mae\"]\n",
        "epochs = range(1, len(loss) + 1)\n",
        "plt.figure()\n",
        "plt.plot(epochs, loss, \"bo\", label=\"Training MAE\")\n",
        "plt.plot(epochs, val_loss, \"b\", label=\"Validation MAE\")\n",
        "plt.title(\"Training and validation MAE\")\n",
        "plt.legend()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nHd3-Y89jbce"
      },
      "source": [
        "### 1D 합성곱 모델 시도해 보기"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:30:16.566194Z",
          "iopub.status.busy": "2022-05-20T13:30:16.565904Z",
          "iopub.status.idle": "2022-05-20T13:35:49.724331Z",
          "shell.execute_reply": "2022-05-20T13:35:49.723349Z"
        },
        "id": "MIodgX8jjbce",
        "outputId": "4a579fa7-9901-4107-a104-4456261907a8",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "819/819 [==============================] - 18s 14ms/step - loss: 24.5388 - mae: 3.8288 - val_loss: 21.1984 - val_mae: 3.6167\n",
            "Epoch 2/10\n",
            "819/819 [==============================] - 11s 14ms/step - loss: 16.2037 - mae: 3.2001 - val_loss: 15.4073 - val_mae: 3.0842\n",
            "Epoch 3/10\n",
            "819/819 [==============================] - 12s 14ms/step - loss: 15.1495 - mae: 3.0930 - val_loss: 14.5760 - val_mae: 2.9930\n",
            "Epoch 4/10\n",
            "819/819 [==============================] - 11s 14ms/step - loss: 14.2915 - mae: 3.0055 - val_loss: 14.9988 - val_mae: 3.0766\n",
            "Epoch 5/10\n",
            "819/819 [==============================] - 12s 14ms/step - loss: 13.6080 - mae: 2.9297 - val_loss: 15.5050 - val_mae: 3.1179\n",
            "Epoch 6/10\n",
            "819/819 [==============================] - 11s 14ms/step - loss: 13.0705 - mae: 2.8721 - val_loss: 20.1465 - val_mae: 3.5950\n",
            "Epoch 7/10\n",
            "819/819 [==============================] - 11s 14ms/step - loss: 12.5662 - mae: 2.8113 - val_loss: 16.2455 - val_mae: 3.1957\n",
            "Epoch 8/10\n",
            "819/819 [==============================] - 11s 14ms/step - loss: 12.1728 - mae: 2.7678 - val_loss: 13.5226 - val_mae: 2.8887\n",
            "Epoch 9/10\n",
            "819/819 [==============================] - 11s 14ms/step - loss: 11.7866 - mae: 2.7214 - val_loss: 15.6385 - val_mae: 3.1415\n",
            "Epoch 10/10\n",
            "819/819 [==============================] - 11s 14ms/step - loss: 11.4573 - mae: 2.6830 - val_loss: 13.9089 - val_mae: 2.9268\n",
            "405/405 [==============================] - 4s 9ms/step - loss: 14.5079 - mae: 3.0026\n",
            "테스트 MAE: 3.00\n"
          ]
        }
      ],
      "source": [
        "inputs = keras.Input(shape=(sequence_length, raw_data.shape[-1]))\n",
        "x = layers.Conv1D(8, 24, activation=\"relu\")(inputs)\n",
        "x = layers.MaxPooling1D(2)(x)\n",
        "x = layers.Conv1D(8, 12, activation=\"relu\")(x)\n",
        "x = layers.MaxPooling1D(2)(x)\n",
        "x = layers.Conv1D(8, 6, activation=\"relu\")(x)\n",
        "x = layers.GlobalAveragePooling1D()(x)\n",
        "outputs = layers.Dense(1)(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "\n",
        "callbacks = [\n",
        "    keras.callbacks.ModelCheckpoint(\"jena_conv.keras\",\n",
        "                                    save_best_only=True)\n",
        "]\n",
        "model.compile(optimizer=\"rmsprop\", loss=\"mse\", metrics=[\"mae\"])\n",
        "history = model.fit(train_dataset,\n",
        "                    epochs=10,\n",
        "                    validation_data=val_dataset,\n",
        "                    callbacks=callbacks)\n",
        "\n",
        "model = keras.models.load_model(\"jena_conv.keras\")\n",
        "print(f\"테스트 MAE: {model.evaluate(test_dataset)[1]:.2f}\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:35:49.728113Z",
          "iopub.status.busy": "2022-05-20T13:35:49.727783Z",
          "iopub.status.idle": "2022-05-20T13:35:49.850064Z",
          "shell.execute_reply": "2022-05-20T13:35:49.849311Z"
        },
        "id": "twXzOJXkjbce",
        "outputId": "565963f0-5f65-420a-cb5e-65a0a1d24bd5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "loss = history.history[\"mae\"]\n",
        "val_loss = history.history[\"val_mae\"]\n",
        "epochs = range(1, len(loss) + 1)\n",
        "plt.figure()\n",
        "plt.plot(epochs, loss, \"bo\", label=\"Training MAE\")\n",
        "plt.plot(epochs, val_loss, \"b\", label=\"Validation MAE\")\n",
        "plt.title(\"Training and validation MAE\")\n",
        "plt.legend()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SIVEXb7ojbcf"
      },
      "source": [
        "### 첫 번째 순환 신경망"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "KGsF8BjGjbcf"
      },
      "source": [
        "**간단한 LSTM 기반 모델**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:35:49.855589Z",
          "iopub.status.busy": "2022-05-20T13:35:49.855249Z",
          "iopub.status.idle": "2022-05-20T13:46:15.333207Z",
          "shell.execute_reply": "2022-05-20T13:46:15.332248Z"
        },
        "id": "ElIY09G2jbcg",
        "outputId": "98467226-bedf-4c94-8353-4e71c9381ea7",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "819/819 [==============================] - 17s 17ms/step - loss: 38.7303 - mae: 4.5164 - val_loss: 12.1018 - val_mae: 2.6434\n",
            "Epoch 2/10\n",
            "819/819 [==============================] - 14s 17ms/step - loss: 10.7679 - mae: 2.5511 - val_loss: 9.8098 - val_mae: 2.4582\n",
            "Epoch 3/10\n",
            "819/819 [==============================] - 14s 17ms/step - loss: 9.7268 - mae: 2.4285 - val_loss: 10.1011 - val_mae: 2.4838\n",
            "Epoch 4/10\n",
            "819/819 [==============================] - 14s 17ms/step - loss: 9.2670 - mae: 2.3683 - val_loss: 10.3945 - val_mae: 2.5106\n",
            "Epoch 5/10\n",
            "819/819 [==============================] - 14s 17ms/step - loss: 8.9701 - mae: 2.3302 - val_loss: 10.1427 - val_mae: 2.4688\n",
            "Epoch 6/10\n",
            "819/819 [==============================] - 14s 17ms/step - loss: 8.7401 - mae: 2.2984 - val_loss: 10.5485 - val_mae: 2.5111\n",
            "Epoch 7/10\n",
            "819/819 [==============================] - 14s 17ms/step - loss: 8.5208 - mae: 2.2710 - val_loss: 10.4418 - val_mae: 2.5053\n",
            "Epoch 8/10\n",
            "819/819 [==============================] - 14s 17ms/step - loss: 8.3236 - mae: 2.2416 - val_loss: 10.9559 - val_mae: 2.5624\n",
            "Epoch 9/10\n",
            "819/819 [==============================] - 14s 17ms/step - loss: 8.1351 - mae: 2.2142 - val_loss: 10.9853 - val_mae: 2.5713\n",
            "Epoch 10/10\n",
            "819/819 [==============================] - 14s 17ms/step - loss: 7.9817 - mae: 2.1940 - val_loss: 10.7646 - val_mae: 2.5554\n",
            "405/405 [==============================] - 4s 9ms/step - loss: 10.8685 - mae: 2.5849\n",
            "테스트 MAE: 2.58\n"
          ]
        }
      ],
      "source": [
        "inputs = keras.Input(shape=(sequence_length, raw_data.shape[-1]))\n",
        "x = layers.LSTM(16)(inputs)\n",
        "outputs = layers.Dense(1)(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "\n",
        "callbacks = [\n",
        "    keras.callbacks.ModelCheckpoint(\"jena_lstm.keras\",\n",
        "                                    save_best_only=True)\n",
        "]\n",
        "model.compile(optimizer=\"rmsprop\", loss=\"mse\", metrics=[\"mae\"])\n",
        "history = model.fit(train_dataset,\n",
        "                    epochs=10,\n",
        "                    validation_data=val_dataset,\n",
        "                    callbacks=callbacks)\n",
        "\n",
        "model = keras.models.load_model(\"jena_lstm.keras\")\n",
        "print(f\"테스트 MAE: {model.evaluate(test_dataset)[1]:.2f}\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:46:15.336897Z",
          "iopub.status.busy": "2022-05-20T13:46:15.336598Z",
          "iopub.status.idle": "2022-05-20T13:46:15.462410Z",
          "shell.execute_reply": "2022-05-20T13:46:15.461466Z"
        },
        "id": "EX3qK89tjbcg",
        "outputId": "85a6d8cc-7f1a-458a-e83e-37522242c58b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "loss = history.history[\"mae\"]\n",
        "val_loss = history.history[\"val_mae\"]\n",
        "epochs = range(1, len(loss) + 1)\n",
        "plt.figure()\n",
        "plt.plot(epochs, loss, \"bo\", label=\"Training MAE\")\n",
        "plt.plot(epochs, val_loss, \"b\", label=\"Validation MAE\")\n",
        "plt.title(\"Training and validation MAE\")\n",
        "plt.legend()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OjfiTkmOjbcg"
      },
      "source": [
        "## 순환 신경망 이해하기"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "11qaJzuKjbcg"
      },
      "source": [
        "**넘파이로 구현한 간단한 RNN**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:46:15.466402Z",
          "iopub.status.busy": "2022-05-20T13:46:15.466102Z",
          "iopub.status.idle": "2022-05-20T13:46:15.474183Z",
          "shell.execute_reply": "2022-05-20T13:46:15.473353Z"
        },
        "id": "LqYsbiAGjbcg"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "timesteps = 100\n",
        "input_features = 32\n",
        "output_features = 64\n",
        "inputs = np.random.random((timesteps, input_features))\n",
        "state_t = np.zeros((output_features,))\n",
        "W = np.random.random((output_features, input_features))\n",
        "U = np.random.random((output_features, output_features))\n",
        "b = np.random.random((output_features,))\n",
        "successive_outputs = []\n",
        "for input_t in inputs:\n",
        "    output_t = np.tanh(np.dot(W, input_t) + np.dot(U, state_t) + b)\n",
        "    successive_outputs.append(output_t)\n",
        "    state_t = output_t\n",
        "final_output_sequence = np.stack(successive_outputs, axis=0)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "m5_JDLW5jbch"
      },
      "source": [
        "### 케라스의 순환 층"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8scRPX8yjbch"
      },
      "source": [
        "**어떤 길이의 시퀀스도 처리할 수 있는 RNN 층**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:46:15.477895Z",
          "iopub.status.busy": "2022-05-20T13:46:15.477617Z",
          "iopub.status.idle": "2022-05-20T13:46:15.532690Z",
          "shell.execute_reply": "2022-05-20T13:46:15.531842Z"
        },
        "id": "RDRvy-SDjbch"
      },
      "outputs": [],
      "source": [
        "num_features = 14\n",
        "inputs = keras.Input(shape=(None, num_features))\n",
        "outputs = layers.SimpleRNN(16)(inputs)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "d_cZi_TVjbch"
      },
      "source": [
        "**마지막 출력 스텝만 반환하는 RNN 층**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:46:15.536872Z",
          "iopub.status.busy": "2022-05-20T13:46:15.536352Z",
          "iopub.status.idle": "2022-05-20T13:46:15.590222Z",
          "shell.execute_reply": "2022-05-20T13:46:15.589305Z"
        },
        "id": "7OpmWg1yjbch",
        "outputId": "c64adf1d-ae20-4ff6-fec7-463326952b9e",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(None, 16)\n"
          ]
        }
      ],
      "source": [
        "num_features = 14\n",
        "steps = 120\n",
        "inputs = keras.Input(shape=(steps, num_features))\n",
        "outputs = layers.SimpleRNN(16, return_sequences=False)(inputs)\n",
        "print(outputs.shape)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MvXPhL0qjbch"
      },
      "source": [
        "**전체 출력 시퀀스를 반환하는 RNN 층**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 21,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:46:15.594315Z",
          "iopub.status.busy": "2022-05-20T13:46:15.593481Z",
          "iopub.status.idle": "2022-05-20T13:46:15.647526Z",
          "shell.execute_reply": "2022-05-20T13:46:15.646394Z"
        },
        "id": "jDKDaTY9jbch",
        "outputId": "e8776893-816f-44bf-e387-b9d9ede3930f",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(None, 120, 16)\n"
          ]
        }
      ],
      "source": [
        "num_features = 14\n",
        "steps = 120\n",
        "inputs = keras.Input(shape=(steps, num_features))\n",
        "outputs = layers.SimpleRNN(16, return_sequences=True)(inputs)\n",
        "print(outputs.shape)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Twps5rVojbch"
      },
      "source": [
        "**스태킹(stacking) RNN 층**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:46:15.651371Z",
          "iopub.status.busy": "2022-05-20T13:46:15.651069Z",
          "iopub.status.idle": "2022-05-20T13:46:15.801889Z",
          "shell.execute_reply": "2022-05-20T13:46:15.801117Z"
        },
        "id": "WCXNh9L6jbci"
      },
      "outputs": [],
      "source": [
        "inputs = keras.Input(shape=(steps, num_features))\n",
        "x = layers.SimpleRNN(16, return_sequences=True)(inputs)\n",
        "x = layers.SimpleRNN(16, return_sequences=True)(x)\n",
        "outputs = layers.SimpleRNN(16)(x)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5UTaHUSfjbci"
      },
      "source": [
        "## 순환 신경망의 고급 사용법"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HfvBqBgLjbci"
      },
      "source": [
        "### 과대적합을 감소하기 위해 순환 드롭아웃 사용하기"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tzGShMPQjbci"
      },
      "source": [
        "**드롭아웃 규제를 적용한 LSTM 모델 훈련하고 평가하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T13:46:15.806506Z",
          "iopub.status.busy": "2022-05-20T13:46:15.805817Z",
          "iopub.status.idle": "2022-05-20T15:04:54.368986Z",
          "shell.execute_reply": "2022-05-20T15:04:54.368048Z"
        },
        "id": "ZDfwJ1Osjbci",
        "outputId": "b674711b-bc40-43cf-990b-4bb0b9ef95ee",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/50\n",
            "819/819 [==============================] - 17s 18ms/step - loss: 27.8302 - mae: 3.8852 - val_loss: 9.9197 - val_mae: 2.4421\n",
            "Epoch 2/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 14.7294 - mae: 2.9779 - val_loss: 10.0012 - val_mae: 2.4524\n",
            "Epoch 3/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 13.9053 - mae: 2.8905 - val_loss: 9.7905 - val_mae: 2.4306\n",
            "Epoch 4/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 13.2728 - mae: 2.8218 - val_loss: 9.2261 - val_mae: 2.3659\n",
            "Epoch 5/50\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 12.7657 - mae: 2.7694 - val_loss: 9.5253 - val_mae: 2.4081\n",
            "Epoch 6/50\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 12.2899 - mae: 2.7184 - val_loss: 9.3455 - val_mae: 2.3839\n",
            "Epoch 7/50\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 11.9756 - mae: 2.6825 - val_loss: 9.7602 - val_mae: 2.4408\n",
            "Epoch 8/50\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 11.5854 - mae: 2.6393 - val_loss: 9.7755 - val_mae: 2.4540\n",
            "Epoch 9/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 11.2994 - mae: 2.6079 - val_loss: 9.7048 - val_mae: 2.4522\n",
            "Epoch 10/50\n",
            "819/819 [==============================] - 16s 19ms/step - loss: 11.0032 - mae: 2.5726 - val_loss: 10.0229 - val_mae: 2.4870\n",
            "Epoch 11/50\n",
            "819/819 [==============================] - 16s 19ms/step - loss: 10.7427 - mae: 2.5426 - val_loss: 10.0071 - val_mae: 2.4741\n",
            "Epoch 12/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 10.5363 - mae: 2.5181 - val_loss: 10.0539 - val_mae: 2.4906\n",
            "Epoch 13/50\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 10.2568 - mae: 2.4868 - val_loss: 9.9774 - val_mae: 2.4615\n",
            "Epoch 14/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 10.1436 - mae: 2.4754 - val_loss: 9.9043 - val_mae: 2.4706\n",
            "Epoch 15/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 9.9126 - mae: 2.4453 - val_loss: 10.8807 - val_mae: 2.5845\n",
            "Epoch 16/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 9.7283 - mae: 2.4228 - val_loss: 10.3049 - val_mae: 2.5093\n",
            "Epoch 17/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 9.6204 - mae: 2.4095 - val_loss: 10.7130 - val_mae: 2.5606\n",
            "Epoch 18/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 9.5127 - mae: 2.3954 - val_loss: 11.0451 - val_mae: 2.5935\n",
            "Epoch 19/50\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 9.3654 - mae: 2.3796 - val_loss: 10.5213 - val_mae: 2.5339\n",
            "Epoch 20/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 9.2348 - mae: 2.3565 - val_loss: 10.8843 - val_mae: 2.5733\n",
            "Epoch 21/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 9.1373 - mae: 2.3477 - val_loss: 10.9190 - val_mae: 2.6010\n",
            "Epoch 22/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 9.0214 - mae: 2.3312 - val_loss: 10.6842 - val_mae: 2.5741\n",
            "Epoch 23/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.9849 - mae: 2.3270 - val_loss: 11.1295 - val_mae: 2.6046\n",
            "Epoch 24/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.8910 - mae: 2.3133 - val_loss: 11.2977 - val_mae: 2.6259\n",
            "Epoch 25/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.7409 - mae: 2.2973 - val_loss: 11.2490 - val_mae: 2.6123\n",
            "Epoch 26/50\n",
            "819/819 [==============================] - 16s 19ms/step - loss: 8.6659 - mae: 2.2830 - val_loss: 11.2884 - val_mae: 2.6373\n",
            "Epoch 27/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.5297 - mae: 2.2668 - val_loss: 11.0510 - val_mae: 2.6097\n",
            "Epoch 28/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.5392 - mae: 2.2664 - val_loss: 11.7980 - val_mae: 2.6818\n",
            "Epoch 29/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.4709 - mae: 2.2589 - val_loss: 11.2271 - val_mae: 2.6211\n",
            "Epoch 30/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.4471 - mae: 2.2508 - val_loss: 11.3540 - val_mae: 2.6489\n",
            "Epoch 31/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.2903 - mae: 2.2339 - val_loss: 11.6046 - val_mae: 2.6694\n",
            "Epoch 32/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.2469 - mae: 2.2261 - val_loss: 11.2413 - val_mae: 2.6303\n",
            "Epoch 33/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.1870 - mae: 2.2144 - val_loss: 11.5340 - val_mae: 2.6596\n",
            "Epoch 34/50\n",
            "819/819 [==============================] - 16s 19ms/step - loss: 8.0745 - mae: 2.2018 - val_loss: 11.4612 - val_mae: 2.6615\n",
            "Epoch 35/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 8.0647 - mae: 2.1971 - val_loss: 11.7437 - val_mae: 2.6923\n",
            "Epoch 36/50\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 7.9914 - mae: 2.1889 - val_loss: 12.0667 - val_mae: 2.7215\n",
            "Epoch 37/50\n",
            "819/819 [==============================] - 16s 19ms/step - loss: 8.0160 - mae: 2.1917 - val_loss: 11.3949 - val_mae: 2.6384\n",
            "Epoch 38/50\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 7.9060 - mae: 2.1770 - val_loss: 11.1587 - val_mae: 2.6152\n",
            "Epoch 39/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 7.8514 - mae: 2.1708 - val_loss: 11.8725 - val_mae: 2.7019\n",
            "Epoch 40/50\n",
            "819/819 [==============================] - 16s 19ms/step - loss: 7.7875 - mae: 2.1616 - val_loss: 11.6716 - val_mae: 2.6688\n",
            "Epoch 41/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 7.6848 - mae: 2.1478 - val_loss: 11.7445 - val_mae: 2.6733\n",
            "Epoch 42/50\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 7.7183 - mae: 2.1510 - val_loss: 11.3980 - val_mae: 2.6499\n",
            "Epoch 43/50\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 7.7207 - mae: 2.1510 - val_loss: 11.5768 - val_mae: 2.6663\n",
            "Epoch 44/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 7.6223 - mae: 2.1346 - val_loss: 11.5386 - val_mae: 2.6669\n",
            "Epoch 45/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 7.5538 - mae: 2.1255 - val_loss: 11.3607 - val_mae: 2.6299\n",
            "Epoch 46/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 7.5608 - mae: 2.1255 - val_loss: 11.7910 - val_mae: 2.6855\n",
            "Epoch 47/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 7.4900 - mae: 2.1152 - val_loss: 11.9744 - val_mae: 2.7175\n",
            "Epoch 48/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 7.5101 - mae: 2.1222 - val_loss: 11.7155 - val_mae: 2.6989\n",
            "Epoch 49/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 7.5093 - mae: 2.1166 - val_loss: 11.3403 - val_mae: 2.6489\n",
            "Epoch 50/50\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 7.4202 - mae: 2.1056 - val_loss: 11.6011 - val_mae: 2.6529\n"
          ]
        }
      ],
      "source": [
        "inputs = keras.Input(shape=(sequence_length, raw_data.shape[-1]))\n",
        "# 훈련 속도를 놓이기 위해 순환 드롭아웃을 제외합니다.\n",
        "#x = layers.LSTM(32, recurrent_dropout=0.25)(inputs)\n",
        "x = layers.LSTM(32)(inputs)\n",
        "x = layers.Dropout(0.5)(x)\n",
        "outputs = layers.Dense(1)(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "\n",
        "callbacks = [\n",
        "    keras.callbacks.ModelCheckpoint(\"jena_lstm_dropout.keras\",\n",
        "                                    save_best_only=True)\n",
        "]\n",
        "model.compile(optimizer=\"rmsprop\", loss=\"mse\", metrics=[\"mae\"])\n",
        "history = model.fit(train_dataset,\n",
        "                    epochs=50,\n",
        "                    validation_data=val_dataset,\n",
        "                    callbacks=callbacks)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 25,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T15:04:54.372845Z",
          "iopub.status.busy": "2022-05-20T15:04:54.372302Z",
          "iopub.status.idle": "2022-05-20T15:04:54.500136Z",
          "shell.execute_reply": "2022-05-20T15:04:54.499227Z"
        },
        "id": "TUqqYeRgjbci",
        "outputId": "34e7b633-87dd-4a5f-dc94-129981414e3f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "loss = history.history[\"mae\"]\n",
        "val_loss = history.history[\"val_mae\"]\n",
        "epochs = range(1, len(loss) + 1)\n",
        "plt.figure()\n",
        "plt.plot(epochs, loss, \"bo\", label=\"Training MAE\")\n",
        "plt.plot(epochs, val_loss, \"b\", label=\"Validation MAE\")\n",
        "plt.title(\"Training and validation MAE\")\n",
        "plt.legend()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 26,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T15:04:54.505555Z",
          "iopub.status.busy": "2022-05-20T15:04:54.504713Z",
          "iopub.status.idle": "2022-05-20T15:04:57.937020Z",
          "shell.execute_reply": "2022-05-20T15:04:57.935916Z"
        },
        "id": "u_-Jap5vjbcj",
        "outputId": "1551c085-1a4a-4b69-dce5-4c421ea76560",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:tensorflow:Layer lstm_3 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n"
          ]
        }
      ],
      "source": [
        "inputs = keras.Input(shape=(sequence_length, num_features))\n",
        "x = layers.LSTM(32, recurrent_dropout=0.2, unroll=True)(inputs)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dIAEo0Ttjbcj"
      },
      "source": [
        "### 스태킹 순환 층"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9AfaRKy2jbcj"
      },
      "source": [
        "**드롭아웃 규제와 스태킹을 적용한 GRU 모델을 훈련하고 평가하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 27,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T15:04:57.941005Z",
          "iopub.status.busy": "2022-05-20T15:04:57.940685Z",
          "iopub.status.idle": "2022-05-20T17:21:17.499207Z",
          "shell.execute_reply": "2022-05-20T17:21:17.498103Z"
        },
        "id": "-v7IqHRsjbcj",
        "outputId": "3050b269-ee5a-42c3-e3fc-409eb4c7ae01",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/50\n",
            "819/819 [==============================] - 23s 24ms/step - loss: 23.8203 - mae: 3.6085 - val_loss: 9.1113 - val_mae: 2.3265\n",
            "Epoch 2/50\n",
            "819/819 [==============================] - 19s 24ms/step - loss: 13.4428 - mae: 2.8387 - val_loss: 8.9302 - val_mae: 2.3205\n",
            "Epoch 3/50\n",
            "819/819 [==============================] - 20s 24ms/step - loss: 12.4980 - mae: 2.7414 - val_loss: 9.3066 - val_mae: 2.3674\n",
            "Epoch 4/50\n",
            "819/819 [==============================] - 20s 24ms/step - loss: 11.7165 - mae: 2.6547 - val_loss: 9.7335 - val_mae: 2.4234\n",
            "Epoch 5/50\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 11.0335 - mae: 2.5730 - val_loss: 9.1506 - val_mae: 2.3421\n",
            "Epoch 6/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 10.4060 - mae: 2.5011 - val_loss: 10.8818 - val_mae: 2.5821\n",
            "Epoch 7/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 9.7572 - mae: 2.4213 - val_loss: 10.0641 - val_mae: 2.4677\n",
            "Epoch 8/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 9.2202 - mae: 2.3523 - val_loss: 10.0242 - val_mae: 2.4493\n",
            "Epoch 9/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 8.6846 - mae: 2.2791 - val_loss: 10.9915 - val_mae: 2.5693\n",
            "Epoch 10/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 8.2045 - mae: 2.2138 - val_loss: 10.7338 - val_mae: 2.5448\n",
            "Epoch 11/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 7.7263 - mae: 2.1487 - val_loss: 11.7899 - val_mae: 2.6686\n",
            "Epoch 12/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 7.3051 - mae: 2.0888 - val_loss: 10.9794 - val_mae: 2.5575\n",
            "Epoch 13/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 6.9801 - mae: 2.0386 - val_loss: 11.7308 - val_mae: 2.6494\n",
            "Epoch 14/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 6.7072 - mae: 1.9937 - val_loss: 12.3358 - val_mae: 2.7208\n",
            "Epoch 15/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 6.4467 - mae: 1.9533 - val_loss: 12.6775 - val_mae: 2.7423\n",
            "Epoch 16/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 6.2607 - mae: 1.9220 - val_loss: 12.3900 - val_mae: 2.7246\n",
            "Epoch 17/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 6.1198 - mae: 1.8976 - val_loss: 12.8926 - val_mae: 2.7709\n",
            "Epoch 18/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 5.9531 - mae: 1.8747 - val_loss: 12.8124 - val_mae: 2.7678\n",
            "Epoch 19/50\n",
            "819/819 [==============================] - 20s 24ms/step - loss: 5.8125 - mae: 1.8464 - val_loss: 12.6254 - val_mae: 2.7310\n",
            "Epoch 20/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 5.6929 - mae: 1.8291 - val_loss: 13.1190 - val_mae: 2.7949\n",
            "Epoch 21/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 5.5364 - mae: 1.8033 - val_loss: 13.0416 - val_mae: 2.7760\n",
            "Epoch 22/50\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 5.4661 - mae: 1.7903 - val_loss: 12.9257 - val_mae: 2.7873\n",
            "Epoch 23/50\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 5.3606 - mae: 1.7717 - val_loss: 13.1139 - val_mae: 2.8072\n",
            "Epoch 24/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 5.2825 - mae: 1.7582 - val_loss: 13.0957 - val_mae: 2.7797\n",
            "Epoch 25/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 5.2180 - mae: 1.7452 - val_loss: 13.5254 - val_mae: 2.8283\n",
            "Epoch 26/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 5.1391 - mae: 1.7301 - val_loss: 13.2821 - val_mae: 2.8258\n",
            "Epoch 27/50\n",
            "819/819 [==============================] - 20s 24ms/step - loss: 5.1092 - mae: 1.7237 - val_loss: 13.6206 - val_mae: 2.8514\n",
            "Epoch 28/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 5.0449 - mae: 1.7138 - val_loss: 13.4178 - val_mae: 2.8327\n",
            "Epoch 29/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 5.0141 - mae: 1.7055 - val_loss: 13.5538 - val_mae: 2.8488\n",
            "Epoch 30/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.9340 - mae: 1.6908 - val_loss: 13.9373 - val_mae: 2.8897\n",
            "Epoch 31/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.8808 - mae: 1.6823 - val_loss: 13.5699 - val_mae: 2.8522\n",
            "Epoch 32/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.8787 - mae: 1.6787 - val_loss: 13.4075 - val_mae: 2.8316\n",
            "Epoch 33/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.7971 - mae: 1.6658 - val_loss: 13.5544 - val_mae: 2.8400\n",
            "Epoch 34/50\n",
            "819/819 [==============================] - 20s 24ms/step - loss: 4.7912 - mae: 1.6637 - val_loss: 13.2974 - val_mae: 2.8338\n",
            "Epoch 35/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.7012 - mae: 1.6478 - val_loss: 13.4067 - val_mae: 2.8359\n",
            "Epoch 36/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.6930 - mae: 1.6452 - val_loss: 13.5607 - val_mae: 2.8645\n",
            "Epoch 37/50\n",
            "819/819 [==============================] - 21s 26ms/step - loss: 4.6691 - mae: 1.6411 - val_loss: 13.6249 - val_mae: 2.8655\n",
            "Epoch 38/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.6260 - mae: 1.6326 - val_loss: 13.1008 - val_mae: 2.7956\n",
            "Epoch 39/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.6349 - mae: 1.6344 - val_loss: 13.4788 - val_mae: 2.8514\n",
            "Epoch 40/50\n",
            "819/819 [==============================] - 21s 26ms/step - loss: 4.5712 - mae: 1.6207 - val_loss: 13.2992 - val_mae: 2.8255\n",
            "Epoch 41/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.5403 - mae: 1.6156 - val_loss: 13.5922 - val_mae: 2.8693\n",
            "Epoch 42/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.5403 - mae: 1.6111 - val_loss: 13.3656 - val_mae: 2.8405\n",
            "Epoch 43/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.4899 - mae: 1.6045 - val_loss: 14.5088 - val_mae: 2.9673\n",
            "Epoch 44/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.4676 - mae: 1.5990 - val_loss: 14.2606 - val_mae: 2.9232\n",
            "Epoch 45/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.4431 - mae: 1.5939 - val_loss: 13.8194 - val_mae: 2.8924\n",
            "Epoch 46/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.4127 - mae: 1.5879 - val_loss: 13.2872 - val_mae: 2.8483\n",
            "Epoch 47/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.3908 - mae: 1.5845 - val_loss: 14.0894 - val_mae: 2.9294\n",
            "Epoch 48/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.3258 - mae: 1.5730 - val_loss: 13.1913 - val_mae: 2.8210\n",
            "Epoch 49/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.3674 - mae: 1.5773 - val_loss: 13.1150 - val_mae: 2.8246\n",
            "Epoch 50/50\n",
            "819/819 [==============================] - 20s 25ms/step - loss: 4.3520 - mae: 1.5743 - val_loss: 13.6808 - val_mae: 2.8795\n",
            "405/405 [==============================] - 6s 12ms/step - loss: 10.0774 - mae: 2.4949\n",
            "테스트 MAE: 2.49\n"
          ]
        }
      ],
      "source": [
        "inputs = keras.Input(shape=(sequence_length, raw_data.shape[-1]))\n",
        "# 훈련 속도를 놓이기 위해 순환 드롭아웃을 제외합니다.\n",
        "# x = layers.GRU(32, recurrent_dropout=0.5, return_sequences=True)(inputs)\n",
        "# x = layers.GRU(32, recurrent_dropout=0.5)(x)\n",
        "x = layers.GRU(32, return_sequences=True)(inputs)\n",
        "x = layers.GRU(32)(x)\n",
        "x = layers.Dropout(0.5)(x)\n",
        "outputs = layers.Dense(1)(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "\n",
        "callbacks = [\n",
        "    keras.callbacks.ModelCheckpoint(\"jena_stacked_gru_dropout.keras\",\n",
        "                                    save_best_only=True)\n",
        "]\n",
        "model.compile(optimizer=\"rmsprop\", loss=\"mse\", metrics=[\"mae\"])\n",
        "history = model.fit(train_dataset,\n",
        "                    epochs=50,\n",
        "                    validation_data=val_dataset,\n",
        "                    callbacks=callbacks)\n",
        "model = keras.models.load_model(\"jena_stacked_gru_dropout.keras\")\n",
        "print(f\"테스트 MAE: {model.evaluate(test_dataset)[1]:.2f}\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 28,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T17:21:17.502935Z",
          "iopub.status.busy": "2022-05-20T17:21:17.502627Z",
          "iopub.status.idle": "2022-05-20T17:21:17.637424Z",
          "shell.execute_reply": "2022-05-20T17:21:17.636229Z"
        },
        "id": "ajG9EuXxjbck",
        "outputId": "493a965e-f19f-4c44-ffeb-c1668a53d05e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "loss = history.history[\"mae\"]\n",
        "val_loss = history.history[\"val_mae\"]\n",
        "epochs = range(1, len(loss) + 1)\n",
        "plt.figure()\n",
        "plt.plot(epochs, loss, \"bo\", label=\"Training MAE\")\n",
        "plt.plot(epochs, val_loss, \"b\", label=\"Validation MAE\")\n",
        "plt.title(\"Training and validation MAE\")\n",
        "plt.legend()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Yj_01R2Cjbck"
      },
      "source": [
        "### 양방향 RNN 사용하기"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HfYzAJbqjbck"
      },
      "source": [
        "시퀀스를 거꾸로 뒤집어 LSTM 모델 훈련하기 (책에는 없음)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 29,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T17:21:17.641176Z",
          "iopub.status.busy": "2022-05-20T17:21:17.640902Z",
          "iopub.status.idle": "2022-05-20T17:21:17.646145Z",
          "shell.execute_reply": "2022-05-20T17:21:17.645090Z"
        },
        "id": "P2Ns9cJajbck"
      },
      "outputs": [],
      "source": [
        "def train_generator():\n",
        "    while True:\n",
        "        for samples, targets in train_dataset:\n",
        "            yield samples[:, ::-1, :], targets\n",
        "\n",
        "def val_generator():\n",
        "    while True:\n",
        "        for samples, targets in val_dataset:\n",
        "            yield samples[:, ::-1, :], targets\n",
        "        \n",
        "train_gen = train_generator()\n",
        "val_gen = val_generator()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 30,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T17:21:17.649630Z",
          "iopub.status.busy": "2022-05-20T17:21:17.649357Z",
          "iopub.status.idle": "2022-05-20T17:38:03.455221Z",
          "shell.execute_reply": "2022-05-20T17:38:03.454271Z"
        },
        "id": "4Yd7xpW3jbck",
        "outputId": "b882ee53-6a78-4e10-ea5e-6b60477a5d0f",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "819/819 [==============================] - 17s 19ms/step - loss: 34.4291 - mae: 4.5078 - val_loss: 18.8545 - val_mae: 3.4016\n",
            "Epoch 2/10\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 22.1811 - mae: 3.6915 - val_loss: 17.3549 - val_mae: 3.2415\n",
            "Epoch 3/10\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 20.1641 - mae: 3.5055 - val_loss: 16.4726 - val_mae: 3.1400\n",
            "Epoch 4/10\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 19.4260 - mae: 3.4325 - val_loss: 15.1944 - val_mae: 3.0632\n",
            "Epoch 5/10\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 16.4530 - mae: 3.1715 - val_loss: 12.5819 - val_mae: 2.7369\n",
            "Epoch 6/10\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 14.9640 - mae: 3.0180 - val_loss: 12.3661 - val_mae: 2.7029\n",
            "Epoch 7/10\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 13.9965 - mae: 2.9188 - val_loss: 13.3859 - val_mae: 2.8222\n",
            "Epoch 8/10\n",
            "819/819 [==============================] - 15s 18ms/step - loss: 13.4201 - mae: 2.8536 - val_loss: 12.0510 - val_mae: 2.6898\n",
            "Epoch 9/10\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 12.8782 - mae: 2.7925 - val_loss: 11.6169 - val_mae: 2.6293\n",
            "Epoch 10/10\n",
            "819/819 [==============================] - 15s 19ms/step - loss: 12.4951 - mae: 2.7453 - val_loss: 11.1915 - val_mae: 2.5769\n"
          ]
        }
      ],
      "source": [
        "inputs = keras.Input(shape=(sequence_length, raw_data.shape[-1]))\n",
        "# 훈련 속도를 놓이기 위해 순환 드롭아웃을 제외합니다.\n",
        "# x = layers.LSTM(32, recurrent_dropout=0.25)(inputs)\n",
        "x = layers.LSTM(32)(inputs)\n",
        "x = layers.Dropout(0.5)(x)\n",
        "outputs = layers.Dense(1)(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "\n",
        "model.compile(optimizer=\"rmsprop\", loss=\"mse\", metrics=[\"mae\"])\n",
        "history = model.fit(train_gen,\n",
        "                    epochs=10,\n",
        "                    steps_per_epoch=819,\n",
        "                    validation_data=val_gen,\n",
        "                    validation_steps=410)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 31,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T17:38:03.459157Z",
          "iopub.status.busy": "2022-05-20T17:38:03.458872Z",
          "iopub.status.idle": "2022-05-20T17:38:03.588800Z",
          "shell.execute_reply": "2022-05-20T17:38:03.587843Z"
        },
        "id": "HjBY66fkjbck",
        "outputId": "5a27fe47-f9ed-4d3e-9db5-92931a785f3b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        }
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "loss = history.history[\"mae\"]\n",
        "val_loss = history.history[\"val_mae\"]\n",
        "epochs = range(1, len(loss) + 1)\n",
        "plt.figure()\n",
        "plt.plot(epochs, loss, \"bo\", label=\"Training MAE\")\n",
        "plt.plot(epochs, val_loss, \"b\", label=\"Validation MAE\")\n",
        "plt.title(\"Training and validation MAE\")\n",
        "plt.legend()\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9mc9iUIKjbck"
      },
      "source": [
        "**양방향 LSTM 모델 훈련하고 평가하기**"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 32,
      "metadata": {
        "execution": {
          "iopub.execute_input": "2022-05-20T17:38:03.594401Z",
          "iopub.status.busy": "2022-05-20T17:38:03.593387Z",
          "iopub.status.idle": "2022-05-20T17:49:14.410464Z",
          "shell.execute_reply": "2022-05-20T17:49:14.409759Z"
        },
        "id": "eysu8EhMjbcl",
        "outputId": "1183e67f-105e-4d01-ab34-94596f8a25b3",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "819/819 [==============================] - 25s 27ms/step - loss: 27.5711 - mae: 3.7582 - val_loss: 10.9142 - val_mae: 2.5547\n",
            "Epoch 2/10\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 9.5186 - mae: 2.4085 - val_loss: 10.1476 - val_mae: 2.4799\n",
            "Epoch 3/10\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 8.5681 - mae: 2.2782 - val_loss: 9.8487 - val_mae: 2.4442\n",
            "Epoch 4/10\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 8.0482 - mae: 2.2116 - val_loss: 10.2152 - val_mae: 2.4823\n",
            "Epoch 5/10\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 7.6664 - mae: 2.1611 - val_loss: 10.3464 - val_mae: 2.4819\n",
            "Epoch 6/10\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 7.3586 - mae: 2.1188 - val_loss: 10.7364 - val_mae: 2.5336\n",
            "Epoch 7/10\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 7.1247 - mae: 2.0830 - val_loss: 10.9647 - val_mae: 2.5570\n",
            "Epoch 8/10\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 6.9400 - mae: 2.0525 - val_loss: 10.9305 - val_mae: 2.5544\n",
            "Epoch 9/10\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 6.7492 - mae: 2.0250 - val_loss: 11.5026 - val_mae: 2.6071\n",
            "Epoch 10/10\n",
            "819/819 [==============================] - 21s 25ms/step - loss: 6.5950 - mae: 1.9994 - val_loss: 11.4990 - val_mae: 2.6209\n"
          ]
        }
      ],
      "source": [
        "inputs = keras.Input(shape=(sequence_length, raw_data.shape[-1]))\n",
        "x = layers.Bidirectional(layers.LSTM(16))(inputs)\n",
        "outputs = layers.Dense(1)(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "\n",
        "model.compile(optimizer=\"rmsprop\", loss=\"mse\", metrics=[\"mae\"])\n",
        "history = model.fit(train_dataset,\n",
        "                    epochs=10,\n",
        "                    validation_data=val_dataset)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wrkE0wkWjbcl"
      },
      "source": [
        "### 더 나아가서"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fi2t0VRNjbcl"
      },
      "source": [
        "## 요약"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [],
      "name": "chapter10_dl-for-timeseries.i",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "default:Python",
      "language": "python",
      "name": "conda-env-default-py"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.9.10"
    },
    "accelerator": "GPU",
    "gpuClass": "premium"
  },
  "nbformat": 4,
  "nbformat_minor": 0
}