Skip to content

Instantly share code, notes, and snippets.

@FinanceData

FinanceData/notebook.ipynb

Last active June 6, 2021 15:02
Show Gist options
  • Save FinanceData/f2ba3b4385754e1596f25ae27fae6efa to your computer and use it in GitHub Desktop.
Save FinanceData/f2ba3b4385754e1596f25ae27fae6efa to your computer and use it in GitHub Desktop.
Download ZIP
Keras 08 - LSTM - 다양한 구성.ipynb
Display the source blob
Display the rendered blob
Raw
notebook.ipynb
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"name": "Keras 08 - LSTM - 다양한 구성.ipynb",
"provenance": [],
"collapsed_sections": [],
"include_colab_link": true
},
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"accelerator": "GPU"
},
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "view-in-github",
"colab_type": "text"
},
"source": [
"<a href=\"https://colab.research.google.com/gist/FinanceData/f2ba3b4385754e1596f25ae27fae6efa/notebook.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "NevktENNyu5d"
},
"source": [
"# Keras - LSTM - 다양한 구성\n",
"\n",
"<img width=\"40%\" src=\"https://imgur.com/ZZUfQco.png\">\n",
"\n",
"**2019-2021 [FinanceData.KR](http://financedata.kr) | [facebook.com/financedata](http://facebook.com/financedata)**"
]
},
{
"cell_type": "code",
"metadata": {
"id": "6l7fJTDHBbIK"
},
"source": [
"import numpy as np\n",
"import pandas as pd"
],
"execution_count": 29,
"outputs": []
},
{
"cell_type": "markdown",
"metadata": {
"id": "e0_soJrrkdF6"
},
"source": [
"# LSTM\n",
"입력 (batch_size, timesteps, input_dim) 형태의 3D 텐서\n",
"\n",
"* units: 출력수\n",
"* activation: 활성함수(기본값: `tanh`)\n",
"* stateful: batch내 마지막 상태를 다음 batch의 초기 값으로 전달 (기본값: False) \n",
"\n",
"\n",
"```python\n",
"model = Sequential()\n",
"model.add(LSTM(128, input_shape=(4, 1)))\n",
"model.add(Dense(one_hot_vec_size, activation='softmax'))\n",
"```\n",
"\n",
"\n",
"batch_input_shape"
]
},
{
"cell_type": "code",
"metadata": {
"id": "wm5TJm3XnpVP",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "45c14783-6168-4ed9-ceb3-e631c3aa0fb0"
},
"source": [
"from tensorflow.keras.layers import LSTM\n",
"\n",
"inputs = np.random.random([10, 5, 8]).astype(np.float32)\n",
"lstm = LSTM(4)\n",
"\n",
"output = lstm(inputs) # The output has shape `[10, 4]`.\n",
"output"
],
"execution_count": 30,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"<tf.Tensor: shape=(10, 4), dtype=float32, numpy=\n",
"array([[ 0.20688331, -0.33569196, -0.26950288, -0.32427326],\n",
" [ 0.23900172, -0.3384517 , -0.3084644 , -0.3237447 ],\n",
" [ 0.20406622, -0.38868618, -0.3222125 , -0.37952414],\n",
" [ 0.13626516, -0.3892799 , -0.18139675, -0.37758175],\n",
" [ 0.24495627, -0.4281433 , -0.28752932, -0.3744105 ],\n",
" [ 0.28998747, -0.29479292, -0.24098292, -0.31863555],\n",
" [ 0.18514459, -0.33741927, -0.26141754, -0.3624317 ],\n",
" [ 0.14366014, -0.34975567, -0.23518541, -0.34933624],\n",
" [ 0.24872026, -0.29057756, -0.23649417, -0.31782204],\n",
" [ 0.18731515, -0.34731048, -0.2634387 , -0.35081065]],\n",
" dtype=float32)>"
]
},
"metadata": {
"tags": []
},
"execution_count": 30
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "UoU3W9_voNx7"
},
"source": [
"from tensorflow.keras.models import Sequential\n",
"from tensorflow.keras.layers import LSTM\n",
"\n",
"lstm = LSTM(4, return_sequences=True, return_state=True)\n",
"\n",
"# whole_sequence_output has shape `[32, 10, 4]`.\n",
"# final_memory_state and final_carry_state both have shape `[32, 4]`.\n",
"whole_sequence_output, final_memory_state, final_carry_state = lstm(inputs)"
],
"execution_count": 31,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "DQSLub0hpdAH",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "04801218-3203-4d8f-b72b-125fcd283771"
},
"source": [
"final_carry_state"
],
"execution_count": 32,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"<tf.Tensor: shape=(10, 4), dtype=float32, numpy=\n",
"array([[-0.23477821, -0.2561686 , -1.1772712 , -0.07495873],\n",
" [-0.32611296, -0.17704253, -1.3611389 , -0.0666235 ],\n",
" [ 0.04244636, -0.45983917, -1.2444476 , -0.24431005],\n",
" [-0.27068228, -0.5197038 , -1.2941011 , -0.23620991],\n",
" [-0.22558141, -0.46686926, -1.1262234 , -0.22825663],\n",
" [-0.12436227, -0.4464874 , -1.1424009 , -0.22110598],\n",
" [-0.07241282, -0.43684483, -1.1626145 , -0.18896072],\n",
" [-0.24880418, -0.46992603, -1.3002108 , -0.27417603],\n",
" [-0.27308583, -0.5517089 , -1.3240043 , -0.3272207 ],\n",
" [-0.04194528, -0.33178303, -1.2448735 , -0.095449 ]],\n",
" dtype=float32)>"
]
},
"metadata": {
"tags": []
},
"execution_count": 32
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "SBcQlUVyjb7U"
},
"source": [
"# LSTM 구성 조합\n",
"n-to-m, Direction, Layered 등에 따라 다양한 조합이 가능하다\n",
"\n",
"1. **n-to-m** : many-to-one, many-to-many, one-to-many\n",
"1. **Direction** : Unidirectional(단방향), Bidirectional (양방향)\n",
"1. **Layered** : Single-layered, Multi-layered (Stacked)\n",
"\n",
"\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "tyA0ZGv7PksA"
},
"source": [
"# many-to-one LSTM"
]
},
{
"cell_type": "code",
"metadata": {
"id": "odA7XzY7QOGV",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "81b1cf29-e93e-4e7f-ae59-7cdc1412c57a"
},
"source": [
"from tensorflow.keras.models import Sequential\n",
"from tensorflow.keras.layers import Dense, LSTM\n",
"\n",
"x = np.array([[[1],[2],[3],[4]]])\n",
"y = np.array([[5]])\n",
"\n",
"model = Sequential()\n",
"model.add(LSTM(8, input_shape=(4, 1)))\n",
"model.add(Dense(1))\n",
"model.compile(loss='mse', optimizer='adam')\n",
"\n",
"model.summary()"
],
"execution_count": 33,
"outputs": [
{
"output_type": "stream",
"text": [
"Model: \"sequential_6\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"lstm_12 (LSTM) (None, 8) 320 \n",
"_________________________________________________________________\n",
"dense_6 (Dense) (None, 1) 9 \n",
"=================================================================\n",
"Total params: 329\n",
"Trainable params: 329\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "eUD2mzZoBlcA",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "4de4d608-d95a-4de1-b730-a2fb68793152"
},
"source": [
"x.shape, y.shape"
],
"execution_count": 34,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"((1, 4, 1), (1, 1))"
]
},
"metadata": {
"tags": []
},
"execution_count": 34
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "FxcuUuJkBNoy",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "56e8b31f-dc48-4d97-d555-3f4237802e8d"
},
"source": [
"hist = model.fit(x, y, epochs=500, batch_size=1, verbose=0)\n",
"model.predict(x, batch_size=1)"
],
"execution_count": 35,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"array([[4.9985213]], dtype=float32)"
]
},
"metadata": {
"tags": []
},
"execution_count": 35
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "zHaSuR7LAkL8",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 282
},
"outputId": "18e35902-9638-444a-cc18-e756221bc460"
},
"source": [
"pd.DataFrame(hist.history).plot()"
],
"execution_count": 36,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"<matplotlib.axes._subplots.AxesSubplot at 0x7f905981c6d0>"
]
},
"metadata": {
"tags": []
},
"execution_count": 36
},
{
"output_type": "display_data",
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"tags": [],
"needs_background": "light"
}
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "aCs8qCiXkSMQ"
},
"source": [
"# many-to-many LSTM "
]
},
{
"cell_type": "code",
"metadata": {
"id": "TsclZBkqRX3m",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "1bf9f239-61d5-461e-c30c-4f14494646f9"
},
"source": [
"from tensorflow.keras.models import Model\n",
"from tensorflow.keras.layers import Dense, LSTM, TimeDistributed\n",
"import numpy as np\n",
"\n",
"x = np.array([[[1], [2], [3], [4]]])\n",
"y = np.array([[[2], [3], [4], [5]]])\n",
"\n",
"model = Sequential()\n",
"model.add(LSTM(8, input_shape=(4, 1), return_sequences=True))\n",
"model.add(TimeDistributed(Dense(1)))\n",
"model.compile(loss='mse', optimizer='adam')\n",
"\n",
"model.summary()"
],
"execution_count": 37,
"outputs": [
{
"output_type": "stream",
"text": [
"Model: \"sequential_7\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"lstm_13 (LSTM) (None, 4, 8) 320 \n",
"_________________________________________________________________\n",
"time_distributed_2 (TimeDist (None, 4, 1) 9 \n",
"=================================================================\n",
"Total params: 329\n",
"Trainable params: 329\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "ejdqLOaFyIKL",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 282
},
"outputId": "37f67653-3031-43b8-e98f-e0677e3ccc3f"
},
"source": [
"hist = model.fit(x, y, epochs=500, batch_size=1, verbose=0)\n",
"\n",
"pd.DataFrame(hist.history).plot()"
],
"execution_count": 38,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"<matplotlib.axes._subplots.AxesSubplot at 0x7f9058224b10>"
]
},
"metadata": {
"tags": []
},
"execution_count": 38
},
{
"output_type": "display_data",
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"tags": [],
"needs_background": "light"
}
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "rVPJ0hYsyLs8",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "a411b549-6902-43c1-e124-74615b56aca7"
},
"source": [
"model.predict(x).flatten()"
],
"execution_count": 39,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"array([1.2686777, 3.0646281, 4.3462524, 4.721526 ], dtype=float32)"
]
},
"metadata": {
"tags": []
},
"execution_count": 39
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "32qyUjLbzAiq"
},
"source": [
"# Bidirectional LSTM\n",
"* 시퀀스 입력값에 대해 이전 데이터와 후 데이터와의 관계도 학습\n",
"* 두 개의 모델(순방향, 역방향)을 만들어 학습 결과를 합치(merge)는 모델\n",
"* Forward 레이어 + Backward 레이어\n",
" * Forward 레이어: 1→2→3→4 시퀀스 학습\n",
" * Backward 레이어: 1←2←3←4 시퀀스 학습\n"
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "hZpYhfXxvk92"
},
"source": [
"# Bidirectional many-to-one LSTM"
]
},
{
"cell_type": "code",
"metadata": {
"id": "kWZZt1lXz21U",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "f8eeb97b-4aae-45e0-bea1-8c2c9fd83410"
},
"source": [
"from tensorflow.keras.models import Sequential\n",
"from tensorflow.keras.layers import Dense, LSTM, Bidirectional\n",
"import numpy as np\n",
"\n",
"x = np.array([[[1], [2], [3], [4]]])\n",
"y = np.array([[5]])\n",
"\n",
"model = Sequential()\n",
"model.add(Bidirectional(LSTM(8), input_shape=(4, 1), merge_mode='concat'))\n",
"model.add(Dense(1))\n",
"\n",
"model.compile(loss='mse', optimizer='adam')\n",
"\n",
"model.summary()"
],
"execution_count": 40,
"outputs": [
{
"output_type": "stream",
"text": [
"Model: \"sequential_8\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"bidirectional_3 (Bidirection (None, 16) 640 \n",
"_________________________________________________________________\n",
"dense_8 (Dense) (None, 1) 17 \n",
"=================================================================\n",
"Total params: 657\n",
"Trainable params: 657\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "7dXa_l7D2Y9k"
},
"source": [
"hist = model.fit(x, y, epochs=500, batch_size=1, verbose=0)"
],
"execution_count": 41,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "petPL7NZwyL7",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 282
},
"outputId": "86215314-faf8-48c0-9dcb-a0f88cf547e2"
},
"source": [
"pd.DataFrame(hist.history).plot()"
],
"execution_count": 42,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"<matplotlib.axes._subplots.AxesSubplot at 0x7f90507d2fd0>"
]
},
"metadata": {
"tags": []
},
"execution_count": 42
},
{
"output_type": "display_data",
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"tags": [],
"needs_background": "light"
}
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "yqLmYeNA2gH_",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "db704eb9-786b-49ba-ac6e-ef175f246cf6"
},
"source": [
"model.predict(x)"
],
"execution_count": 43,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"array([[5.000004]], dtype=float32)"
]
},
"metadata": {
"tags": []
},
"execution_count": 43
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "PoITN2an20lS"
},
"source": [
"# Bidirectional many-to-many LSTM\n",
"* LSTM return_sequences=True + Bidirectional\n",
"* TimeDistributed()\n"
]
},
{
"cell_type": "code",
"metadata": {
"id": "m3eHau213WmI",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "90b5b037-7228-45af-cbf6-76893c707a0a"
},
"source": [
"from tensorflow.keras.models import Sequential\n",
"from tensorflow.keras.layers import Dense, LSTM, Bidirectional\n",
"import numpy as np\n",
"\n",
"x = np.array([[[1], [2], [3], [4]]])\n",
"y = np.array([[[2], [3], [4], [5]]])\n",
"\n",
"model = Sequential()\n",
"model.add(Bidirectional(LSTM(8, return_sequences=True), input_shape=(4, 1), merge_mode='concat'))\n",
"model.add(TimeDistributed(Dense(1)))\n",
"\n",
"model.compile(loss='mse', optimizer='adam')\n",
"\n",
"model.summary()"
],
"execution_count": 44,
"outputs": [
{
"output_type": "stream",
"text": [
"Model: \"sequential_9\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"bidirectional_4 (Bidirection (None, 4, 16) 640 \n",
"_________________________________________________________________\n",
"time_distributed_3 (TimeDist (None, 4, 1) 17 \n",
"=================================================================\n",
"Total params: 657\n",
"Trainable params: 657\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "hzuF1Ajeweky"
},
"source": [
"hist = model.fit(x, y, epochs=500, batch_size=1, verbose=0)"
],
"execution_count": 45,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "TWrm1TzCwhJ9",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 282
},
"outputId": "2e43176c-f6b9-406e-8b53-f826cd720db0"
},
"source": [
"pd.DataFrame(hist.history).plot()"
],
"execution_count": 46,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"<matplotlib.axes._subplots.AxesSubplot at 0x7f905bf0a5d0>"
]
},
"metadata": {
"tags": []
},
"execution_count": 46
},
{
"output_type": "display_data",
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"tags": [],
"needs_background": "light"
}
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "CjlD5V7HwhXl",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "03d8befd-625d-4e66-b1a4-c49010a0d1c6"
},
"source": [
"model.predict(x)"
],
"execution_count": 47,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"array([[[1.9752381],\n",
" [3.066146 ],\n",
" [4.2643476],\n",
" [4.6654572]]], dtype=float32)"
]
},
"metadata": {
"tags": []
},
"execution_count": 47
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "wxKqj_3dwz7d"
},
"source": [
"# Stacked many-to-one LSTM\n",
"다음 레이어가 LSTM 레이어라면 return_sequences=True 를 지정한다 (모든 출력이 각 스텝으로 전달되어야 하기 때문)\n"
]
},
{
"cell_type": "code",
"metadata": {
"id": "gRHtFY6nxbFn",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "12c981ee-3d5f-4527-df84-7d521d8b5b10"
},
"source": [
"from tensorflow.keras.models import Sequential\n",
"from tensorflow.keras.layers import Dense, LSTM\n",
"\n",
"x = np.array([[[1],[2],[3],[4]]])\n",
"y = np.array([[5]])\n",
"\n",
"model = Sequential()\n",
"model.add(LSTM(8, input_shape=(4, 1), return_sequences=True))\n",
"model.add(LSTM(8))\n",
"model.add(Dense(1))\n",
"model.compile(loss='mse', optimizer='adam')\n",
"\n",
"model.summary()"
],
"execution_count": 48,
"outputs": [
{
"output_type": "stream",
"text": [
"Model: \"sequential_10\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"lstm_16 (LSTM) (None, 4, 8) 320 \n",
"_________________________________________________________________\n",
"lstm_17 (LSTM) (None, 8) 544 \n",
"_________________________________________________________________\n",
"dense_10 (Dense) (None, 1) 9 \n",
"=================================================================\n",
"Total params: 873\n",
"Trainable params: 873\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "iZ8oc5iLHjZ9"
},
"source": [
"hist = model.fit(x, y, epochs=500, batch_size=1, verbose=0)"
],
"execution_count": 49,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "5aLpSxqyHjfh",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 282
},
"outputId": "c74298b2-8482-4ea0-f037-cb73f55e14fd"
},
"source": [
"pd.DataFrame(hist.history).plot()"
],
"execution_count": 50,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"<matplotlib.axes._subplots.AxesSubplot at 0x7f905a1c1450>"
]
},
"metadata": {
"tags": []
},
"execution_count": 50
},
{
"output_type": "display_data",
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"tags": [],
"needs_background": "light"
}
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "NYpCmNq7Hmfk",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "31f3a0fe-b6cc-4ed2-de92-0a408a75039f"
},
"source": [
"model.predict(x)"
],
"execution_count": 51,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"array([[4.98183]], dtype=float32)"
]
},
"metadata": {
"tags": []
},
"execution_count": 51
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "IQtltsHEHyR_"
},
"source": [
"# Bidirectional Stacked many-to-one LSTM\n",
"Bidirectional을 Stacked LSTM에 추가\n"
]
},
{
"cell_type": "code",
"metadata": {
"id": "sP2hAjFqHzU-",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "393b34d3-c1ae-488f-f68f-e91f6d96df12"
},
"source": [
"from tensorflow.keras.models import Sequential\n",
"from tensorflow.keras.layers import Dense, LSTM\n",
"\n",
"x = np.array([[[1],[2],[3],[4]]])\n",
"y = np.array([[5]])\n",
"\n",
"model = Sequential()\n",
"model.add(LSTM(8, input_shape=(4, 1), return_sequences=True))\n",
"model.add(Bidirectional(LSTM(8)))\n",
"model.add(Dense(1))\n",
"model.compile(loss='mse', optimizer='adam')\n",
"\n",
"model.summary()"
],
"execution_count": 52,
"outputs": [
{
"output_type": "stream",
"text": [
"Model: \"sequential_11\"\n",
"_________________________________________________________________\n",
"Layer (type) Output Shape Param # \n",
"=================================================================\n",
"lstm_18 (LSTM) (None, 4, 8) 320 \n",
"_________________________________________________________________\n",
"bidirectional_5 (Bidirection (None, 16) 1088 \n",
"_________________________________________________________________\n",
"dense_11 (Dense) (None, 1) 17 \n",
"=================================================================\n",
"Total params: 1,425\n",
"Trainable params: 1,425\n",
"Non-trainable params: 0\n",
"_________________________________________________________________\n"
],
"name": "stdout"
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "MzN9Xy-gHzVD"
},
"source": [
"hist = model.fit(x, y, epochs=500, batch_size=1, verbose=0)"
],
"execution_count": 53,
"outputs": []
},
{
"cell_type": "code",
"metadata": {
"id": "rGRUrzZrHzVF",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 282
},
"outputId": "0d2523e2-31c2-4ae4-d5b6-c5e26109a64f"
},
"source": [
"pd.DataFrame(hist.history).plot()"
],
"execution_count": 54,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"<matplotlib.axes._subplots.AxesSubplot at 0x7f905a1ef510>"
]
},
"metadata": {
"tags": []
},
"execution_count": 54
},
{
"output_type": "display_data",
"data": {
"image/png": "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\n",
"text/plain": [
"<Figure size 432x288 with 1 Axes>"
]
},
"metadata": {
"tags": [],
"needs_background": "light"
}
}
]
},
{
"cell_type": "code",
"metadata": {
"id": "R8mI2uMtHzVI",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 0
},
"outputId": "706354cb-69f8-4672-debe-335552a3026e"
},
"source": [
"model.predict(x)"
],
"execution_count": 55,
"outputs": [
{
"output_type": "execute_result",
"data": {
"text/plain": [
"array([[5.0000005]], dtype=float32)"
]
},
"metadata": {
"tags": []
},
"execution_count": 55
}
]
},
{
"cell_type": "markdown",
"metadata": {
"id": "SbbRBFy9IGOy"
},
"source": [
"**2019-2021 [FinanceData.KR](http://financedata.kr) | [facebook.com/financedata](http://facebook.com/financedata)**"
]
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment