{"id":29233,"date":"2024-10-27T16:17:25","date_gmt":"2024-10-27T16:17:25","guid":{"rendered":"http:\/\/atmokpo.com\/w\/?p=29233"},"modified":"2024-11-26T06:53:32","modified_gmt":"2024-11-26T06:53:32","slug":"%eb%a8%b8%ec%8b%a0%eb%9f%ac%eb%8b%9d-%eb%b0%8f-%eb%94%a5%eb%9f%ac%eb%8b%9d-%ec%95%8c%ea%b3%a0%eb%a6%ac%ec%a6%98-%ed%8a%b8%eb%a0%88%ec%9d%b4%eb%94%a9-%ec%9d%b4%eb%8f%99-%ed%8f%89%ea%b7%a0-%eb%aa%a8","status":"publish","type":"post","link":"https:\/\/atmokpo.com\/w\/29233\/","title":{"rendered":"\uba38\uc2e0\ub7ec\ub2dd \ubc0f \ub525\ub7ec\ub2dd \uc54c\uace0\ub9ac\uc998 \ud2b8\ub808\uc774\ub529, \uc774\ub3d9 \ud3c9\uade0 \ubaa8\ub378\uc758 \uad6c\ucd95"},"content":{"rendered":"

<\/p>\n

\ud604\ub300 \uae08\uc735 \uc2dc\uc7a5\uc5d0\uc11c \uc54c\uace0\ub9ac\uc998 \ud2b8\ub808\uc774\ub529\uc740 \ub9ce\uc740 \ud22c\uc790\uc790\uc640 \ud2b8\ub808\uc774\ub354\uc5d0\uac8c \ud544\uc218\uc801\uc778 \ub3c4\uad6c\uac00 \ub418\uc5c8\uc2b5\ub2c8\ub2e4. \ud2b9\ud788 \uba38\uc2e0\ub7ec\ub2dd(ML) \ubc0f \ub525\ub7ec\ub2dd(DL) \uae30\uc220\uc744 \ud65c\uc6a9\ud55c \uc790\ub3d9\ub9e4\ub9e4 \uc2dc\uc2a4\ud15c\uc740 \uadf8 \ud6a8\uc728\uc131\uacfc \uc815\ud655\uc131 \ub355\ubd84\uc5d0 \uac01\uad11\ubc1b\uace0 \uc788\uc2b5\ub2c8\ub2e4. \uc774 \uac15\uc88c\uc5d0\uc11c\ub294 \uc774\ub3d9 \ud3c9\uade0(MA) \ubaa8\ub378\uc744 \uc911\uc2ec\uc73c\ub85c \uba38\uc2e0\ub7ec\ub2dd \ubc0f \ub525\ub7ec\ub2dd \uc54c\uace0\ub9ac\uc998\uc744 \ud65c\uc6a9\ud55c \ud2b8\ub808\uc774\ub529 \uc2dc\uc2a4\ud15c\uc744 \uad6c\ucd95\ud558\ub294 \ubc29\ubc95\uc5d0 \ub300\ud574 \uc790\uc138\ud788 \ub2e4\ub8e8\uaca0\uc2b5\ub2c8\ub2e4.<\/p>\n

1. \uc774\ub3d9 \ud3c9\uade0(MA) \uac1c\uc694<\/h2>\n

\uc774\ub3d9 \ud3c9\uade0\uc740 \uc8fc\uc2dd, \uc0c1\ud488, \uc678\ud658 \ub4f1 \ub2e4\uc591\ud55c \uc790\uc0b0\uc758 \uac00\uaca9 \ucd94\uc138\ub97c \ubd84\uc11d\ud558\ub294 \ub370 \uc0ac\uc6a9\ub418\ub294 \uae30\ubc95\uc785\ub2c8\ub2e4. \uc8fc\uc2dd\uc758 \uac00\uaca9 \ubcc0\ub3d9\uc131\uc744 \uac10\uc18c\uc2dc\ud0a4\uace0, \uc7a5\uae30\uc801\uc778 \ud2b8\ub80c\ub4dc\ub97c \ud30c\uc545\ud558\uae30 \uc704\ud574 \uacc4\uc0b0\ub429\ub2c8\ub2e4. \uc774\ub3d9 \ud3c9\uade0\uc5d0\ub294 \uc5ec\ub7ec \uc885\ub958\uac00 \uc788\uc73c\uba70, \uadf8 \uc911\uc5d0\uc11c\ub3c4 \uac00\uc7a5 \uc77c\ubc18\uc801\uc73c\ub85c \uc0ac\uc6a9\ub418\ub294 \ub450 \uac00\uc9c0\ub294 \uac04\ub2e8 \uc774\ub3d9 \ud3c9\uade0(SMA)\uacfc \uc9c0\uc218 \uc774\ub3d9 \ud3c9\uade0(EMA)\uc785\ub2c8\ub2e4.<\/p>\n

1.1 \uac04\ub2e8 \uc774\ub3d9 \ud3c9\uade0(SMA)<\/h3>\n

SMA\ub294 \ud2b9\uc815 \uae30\uac04\uc758 \uac00\uaca9\uc744 \ub2e8\uc21c\ud788 \ud3c9\uade0\ud558\uc5ec \uacc4\uc0b0\ud55c \uac12\uc785\ub2c8\ub2e4. \uc608\ub97c \ub4e4\uc5b4, 5\uc77c SMA\ub294 \ucd5c\uadfc 5\uc77c\uac04\uc758 \uc885\uac00\ub97c \ubaa8\ub450 \ub354\ud55c \ud6c4 5\ub85c \ub098\ub208 \uac12\uc785\ub2c8\ub2e4. SMA\ub294 \uc9c1\uad00\uc801\uc774\uba70 \uc774\ud574\ud558\uae30 \uc27d\uc9c0\ub9cc, \uac00\uaca9 \ubcc0\ud654\uc5d0 \ub300\ud574 \ubbfc\uac10\ud558\uc9c0 \uc54a\uc740 \ub2e8\uc810\uc774 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n

1.2 \uc9c0\uc218 \uc774\ub3d9 \ud3c9\uade0(EMA)<\/h3>\n

EMA\ub294 \ucd5c\uadfc \uac00\uaca9\uc5d0 \ub354 \ub9ce\uc740 \uac00\uc911\uce58\ub97c \ubd80\uc5ec\ud558\uc5ec \uacc4\uc0b0\ub418\ubbc0\ub85c \ucd5c\uc2e0 \uac00\uaca9 \ubcc0\ub3d9\uc5d0 \ub354 \ubbfc\uac10\ud569\ub2c8\ub2e4. \uc774\ub294 \ube60\ub974\uac8c \ubcc0\ud558\ub294 \uc2dc\uc7a5\uc5d0\uc11c \ub354\uc6b1 \ud6a8\uacfc\uc801\uc778 \uc9c0\ud45c\uac00 \ub429\ub2c8\ub2e4. EMA\ub294 \ub2e4\uc74c\uacfc \uac19\uc740 \uacf5\uc2dd\uc744 \uc0ac\uc6a9\ud558\uc5ec \uacc4\uc0b0\ub429\ub2c8\ub2e4:<\/p>\n

EMA = (\ud604\uc7ac \uac00\uaca9 * k) + (\uc774\uc804 EMA * (1 - k))\nk = 2 \/ (N + 1)  \/\/ N\uc740 \uc774\ub3d9 \ud3c9\uade0\uc744 \uacc4\uc0b0\ud560 \uae30\uac04\n<\/code><\/pre>\n
2. \uba38\uc2e0\ub7ec\ub2dd\uc744 \ud65c\uc6a9\ud55c \uc774\ub3d9 \ud3c9\uade0 \ubaa8\ub378 \uad6c\ucd95<\/h2>\n
\uba38\uc2e0\ub7ec\ub2dd\uc744 \uc801\uc6a9\ud558\ub294 \uc774\ub3d9 \ud3c9\uade0 \ubaa8\ub378\uc740 \uacfc\uac70 \ub370\uc774\ud130\ub97c \uae30\ubc18\uc73c\ub85c \uc8fc\uc2dd\uc758 \ubbf8\ub798 \uac00\uaca9\uc744 \uc608\uce21\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \ub2e4\uc74c \ub2e8\uacc4\ub85c \uc774\ubc88 \ud504\ub85c\uc81d\ud2b8\uc5d0\uc11c \uc0ac\uc6a9\ud560 \ub370\uc774\ud130\uc14b\uc744 \uc900\ube44\ud558\uace0, \uba38\uc2e0\ub7ec\ub2dd \uc54c\uace0\ub9ac\uc998\uc744 \uc120\ud0dd\ud558\uc5ec \ubaa8\ub378\uc744 \uad6c\ucd95\ud558\ub294 \ubc29\ubc95\uc5d0 \ub300\ud574 \uc54c\uc544\ubcf4\uaca0\uc2b5\ub2c8\ub2e4.<\/p>\n
2.1 \ub370\uc774\ud130 \uc900\ube44<\/h3>\n
\uc8fc\uc2dd \ub370\uc774\ud130\uac00 \ud3ec\ud568\ub41c CSV \ud30c\uc77c\uc744 \uc0ac\uc6a9\ud558\uc5ec \ubaa8\ub378\uc744 \uad6c\ucd95\ud569\ub2c8\ub2e4. \uc77c\ubc18\uc801\uc73c\ub85c \uc8fc\uc2dd \ub370\uc774\ud130\ub294 Open, High, Low, Close, Volume \ub4f1\uc758 \uc5f4\ub85c \uad6c\uc131\ub429\ub2c8\ub2e4. \uc774 \ub370\uc774\ud130\ub97c pandas \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \uc0ac\uc6a9\ud558\uc5ec \ubd88\ub7ec\uc635\ub2c8\ub2e4:<\/p>\n
import pandas as pd\n\n# \ub370\uc774\ud130 \ubd88\ub7ec\uc624\uae30\ndata = pd.read_csv('stock_data.csv')\nprint(data.head())\n<\/code><\/pre>\n
2.2 \ub370\uc774\ud130 \uc804\ucc98\ub9ac<\/h3>\n
\ub370\uc774\ud130 \uc804\ucc98\ub9ac\ub294 \uba38\uc2e0\ub7ec\ub2dd \ubaa8\ub378\uc774 \ub354 \uc798 \ud559\uc2b5\ud560 \uc218 \uc788\ub3c4\ub85d \ud558\ub294 \uc911\uc694\ud55c \ub2e8\uacc4\uc785\ub2c8\ub2e4. \uacb0\uce21\uac12 \ucc98\ub9ac, \uc774\uc0c1\uce58 \uc81c\uac70, \ud2b9\uc131 \uc120\ud0dd \ubc0f \uc2a4\ucf00\uc77c\ub9c1 \ub4f1\uc758 \uc791\uc5c5\uc774 \ud3ec\ud568\ub429\ub2c8\ub2e4. \ud2b9\ud788, \uc774\ub3d9 \ud3c9\uade0\uc744 \uacc4\uc0b0\ud558\uae30 \uc704\ud55c \uc0c8\ub85c\uc6b4 \uc5f4\uc744 \ucd94\uac00\ud574\uc57c \ud569\ub2c8\ub2e4:<\/p>\n
# \uacb0\uce21\uac12 \ucc98\ub9ac\ndata.fillna(method='ffill', inplace=True)\n\n# \uc774\ub3d9 \ud3c9\uade0 \uc5f4 \ucd94\uac00\ndata['SMA_5'] = data['Close'].rolling(window=5).mean()\ndata['EMA_5'] = data['Close'].ewm(span=5, adjust=False).mean()\n<\/code><\/pre>\n
2.3 \ud2b9\uc131\uacfc \ud0c0\uac9f \ubcc0\uc218 \uc124\uc815<\/h3>\n
\uba38\uc2e0\ub7ec\ub2dd \ubaa8\ub378\uc744 \ud6c8\ub828\uc2dc\ud0a4\uae30 \uc704\ud574 \ub370\uc774\ud130\uc758 \uc785\ub825 \ud2b9\uc131\uacfc \uc608\uce21\ud560 \ud0c0\uac9f \ubcc0\uc218\ub97c \uc124\uc815\ud574\uc57c \ud569\ub2c8\ub2e4. \uc608\ub97c \ub4e4\uc5b4, ‘Close’ \uac00\uaca9\uc5d0 \ub300\ud55c \uc608\uce21\uc744 \uc9c4\ud589\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4:<\/p>\n
X = data[['SMA_5', 'EMA_5', 'Volume']]\ny = data['Close'].shift(-1)  # \ub2e4\uc74c \ub0a0\uc758 \uc885\uac00 \uc608\uce21\nX = X[:-1]  # \ub9c8\uc9c0\ub9c9 \ud589 \uc81c\uac70\ny = y[:-1]\n<\/code><\/pre>\n
2.4 \uba38\uc2e0\ub7ec\ub2dd \ubaa8\ub378 \uc120\ud0dd<\/h3>\n
\uc5ec\ub7ec \uba38\uc2e0\ub7ec\ub2dd \uc54c\uace0\ub9ac\uc998 \uc911\uc5d0\uc11c Decision Tree, Random Forest, XGBoost \ub4f1\uc758 \ubaa8\ub378\uc744 \uc120\ud0dd\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uc5ec\uae30\uc11c\ub294 Random Forest\ub97c \uc608\uc2dc\ub85c \ub4e4\uc5b4\ubcf4\uaca0\uc2b5\ub2c8\ub2e4:<\/p>\n
from sklearn.ensemble import RandomForestRegressor\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import mean_squared_error\n\n# \ub370\uc774\ud130 \ubd84\ud560\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)\n\n# \ubaa8\ub378 \ud6c8\ub828\nmodel = RandomForestRegressor(n_estimators=100)\nmodel.fit(X_train, y_train)\n\n# \uc608\uce21\ny_pred = model.predict(X_test)\n\n# \uc131\ub2a5 \ud3c9\uac00\nmse = mean_squared_error(y_test, y_pred)\nprint(f'Mean Squared Error: {mse}')\n<\/code><\/pre>\n
3. \ub525\ub7ec\ub2dd\uc744 \ud65c\uc6a9\ud55c \uc774\ub3d9 \ud3c9\uade0 \ubaa8\ub378 \uad6c\ucd95<\/h2>\n
\ub525\ub7ec\ub2dd\uc744 \ud65c\uc6a9\ud558\uc5ec \ub354 \ubcf5\uc7a1\ud55c \ud328\ud134\uc744 \ud559\uc2b5\ud558\ub294 \uc774\ub3d9 \ud3c9\uade0 \ubaa8\ub378\uc744 \ub9cc\ub4e4\uc5b4\ubcf4\uaca0\uc2b5\ub2c8\ub2e4. TensorFlow\uc640 Keras \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \uc0ac\uc6a9\ud558\uc5ec \uac04\ub2e8\ud55c \uc778\uacf5\uc2e0\uacbd\ub9dd(ANN)\uc744 \uad6c\ud604\ud574\ubcf4\uaca0\uc2b5\ub2c8\ub2e4.<\/p>\n
3.1 \ub370\uc774\ud130 \uc900\ube44 \ubc0f \uc804\ucc98\ub9ac<\/h3>\n
\ub370\uc774\ud130\ub294 \uba38\uc2e0\ub7ec\ub2dd\uacfc \uc720\uc0ac\ud558\uac8c \uc900\ube44\ud558\uc9c0\ub9cc, \ub525\ub7ec\ub2dd \ubaa8\ub378\uc758 \uacbd\uc6b0 \ub9ce\uc740 \ub370\uc774\ud130\uac00 \ud544\uc694\ud558\ubbc0\ub85c \uc880 \ub354 \uae34 \uae30\uac04\uc758 \ub370\uc774\ud130\ub97c \uc0ac\uc6a9\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \ub610\ud55c, \uc2e0\uacbd\ub9dd\uc758 \uc785\ub825\uc740 3D \ud615\uc0c1\uc774\uc5b4\uc57c \ud558\ubbc0\ub85c reshape \uc791\uc5c5\uc774 \ud544\uc694\ud569\ub2c8\ub2e4:<\/p>\n
import numpy as np\n\n# \ub370\uc774\ud130 \uc2a4\ucf00\uc77c\ub9c1\nfrom sklearn.preprocessing import MinMaxScaler\n\nscaler = MinMaxScaler()\nscaled_data = scaler.fit_transform(data[['Close', 'SMA_5', 'EMA_5', 'Volume']])\n\n# \ub370\uc774\ud130 reshape\nX = []\ny = []\nfor i in range(60, len(scaled_data)):\n    X.append(scaled_data[i-60:i])  # 60\uc77c \ub370\uc774\ud130\ub97c \uc785\ub825\uc73c\ub85c \uc0ac\uc6a9\n    y.append(scaled_data[i, 0])   # \uc608\uce21\ud560 \uac12\uc740 \uc885\uac00\nX, y = np.array(X), np.array(y)\n<\/code><\/pre>\n
3.2 ANN \ubaa8\ub378 \uad6c\ucd95<\/h3>\n
Keras\ub97c \uc0ac\uc6a9\ud558\uc5ec \uc778\uacf5\uc2e0\uacbd\ub9dd \ubaa8\ub378\uc744 \uad6c\ucd95\ud569\ub2c8\ub2e4. \uc5ec\uae30\uc11c\ub294 Dense \ub808\uc774\uc5b4\ub97c \ud65c\uc6a9\ud55c \uac04\ub2e8\ud55c \uad6c\uc870\ub97c \uc0ac\uc6a9\ud558\uaca0\uc2b5\ub2c8\ub2e4:<\/p>\n
from keras.models import Sequential\nfrom keras.layers import Dense, LSTM, Dropout\n\nmodel = Sequential()\nmodel.add(LSTM(50, return_sequences=True, input_shape=(X.shape[1], X.shape[2])))\nmodel.add(Dropout(0.2))\nmodel.add(LSTM(50, return_sequences=False))\nmodel.add(Dropout(0.2))\nmodel.add(Dense(25))\nmodel.add(Dense(1))  # \uc608\uce21\ud560 \uc885\uac00\n\n# \ubaa8\ub378 \ucef4\ud30c\uc77c\nmodel.compile(optimizer='adam', loss='mean_squared_error')\n\n# \ubaa8\ub378 \ud6c8\ub828\nmodel.fit(X, y, batch_size=32, epochs=50)\n<\/code><\/pre>\n
3.3 \uc608\uce21 \ubc0f \uc131\ub2a5 \ud3c9\uac00<\/h3>\n
\ud6c8\ub828\ub41c \ubaa8\ub378\uc744 \uc0ac\uc6a9\ud558\uc5ec \uc608\uce21\uc744 \uc218\ud589\ud558\uace0, \uc131\ub2a5 \ud3c9\uac00\ub97c \uc704\ud574 \uc2e4\uc81c \uc885\uac00\uc640 \uc608\uce21 \uc885\uac00\ub97c \ube44\uad50\ud569\ub2c8\ub2e4:<\/p>\n
# \uc608\uce21\npredictions = model.predict(X)\n\n# \uc5ed\uc815\uaddc\ud654\npredictions = scaler.inverse_transform(predictions)\n\n# \uc131\ub2a5 \ud3c9\uac00\nimport matplotlib.pyplot as plt\n\nplt.plot(data['Close'].values, color='blue', label='\uc2e4\uc81c \uc885\uac00')\nplt.plot(range(60, len(predictions) + 60), predictions, color='red', label='\uc608\uce21 \uc885\uac00')\nplt.legend()\nplt.show()\n<\/code><\/pre>\n
4. \uacb0\uacfc \ud574\uc11d \ubc0f \uac1c\uc120 \ubc29\uc548<\/h2>\n
\ubaa8\ub378\uc744 \ud1b5\ud574 \uc5bb\uc740 \uc608\uce21 \uacb0\uacfc\ub97c \ud574\uc11d\ud558\ub294 \uac83\ub3c4 \uc911\uc694\ud55c \uc791\uc5c5\uc785\ub2c8\ub2e4. \uc608\uce21\uc774 \uc2e4\uc81c \uac00\uaca9\uacfc \ube44\uc2b7\ud560\uc218\ub85d \ubaa8\ub378\uc758 \uc131\ub2a5\uc774 \uc88b\uc740 \uac83\uc785\ub2c8\ub2e4. \uc608\uce21 \uacb0\uacfc\uac00 \ubd80\uc815\ud655\ud558\ub2e4\uba74, \ub2e4\uc74c\uacfc \uac19\uc740 \uac1c\uc120 \ubc29\uc548\uc744 \uace0\ub824\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4:<\/p>\n