\uc624\ub298\ub0a0 \uae08\uc735\uc2dc\uc7a5\uc5d0\uc11c \uba38\uc2e0\ub7ec\ub2dd\uacfc \ub525\ub7ec\ub2dd \uae30\uc220\uc740 \ub354\uc6b1 \ubcf4\ud3b8\ud654\ub418\uace0 \uc788\uc73c\uba70, \uc54c\uace0\ub9ac\uc998 \ud2b8\ub808\uc774\ub529(Algorithmic Trading) \ubd84\uc57c\uc5d0\uc11c\ub3c4 \uadf8 \ud65c\uc6a9\ub3c4\uac00 \uc99d\uac00\ud558\uace0 \uc788\uc2b5\ub2c8\ub2e4. \uc774 \uac15\uc88c\uc5d0\uc11c\ub294 \uba38\uc2e0\ub7ec\ub2dd \ubc0f \ub525\ub7ec\ub2dd \uae30\ubc18\uc758 \ud2b8\ub808\uc774\ub529 \uc54c\uace0\ub9ac\uc998\uc744 \uad6c\ud604\ud558\ub294 \ubc29\ubc95\uacfc OpenAI Gym\uc744 \uc774\uc6a9\ud55c \uc2ec\uce35 \uac15\ud654\ud559\uc2b5\uc758 \uac1c\ub150\uacfc \uc801\uc6a9 \ubc29\ubc95\uc5d0 \ub300\ud574 \uc790\uc138\ud788 \uc124\uba85\ud558\uaca0\uc2b5\ub2c8\ub2e4.<\/p>\n
\uc54c\uace0\ub9ac\uc998 \ud2b8\ub808\uc774\ub529\uc774\ub780 \ucef4\ud4e8\ud130 \ud504\ub85c\uadf8\ub7a8\uc744 \uc0ac\uc6a9\ud558\uc5ec \uc0ac\uc804\uc5d0 \uc815\uc758\ub41c \uaddc\uce59\uc5d0 \ub530\ub77c \uae08\uc735\uc0c1\ud488\uc744 \ub9e4\ub9e4\ud558\ub294 \ubc29\uc2dd\uc785\ub2c8\ub2e4. \uc774\ub7ec\ud55c \ud504\ub85c\uadf8\ub7a8\uc740 \uc2dc\uc7a5\uc758 \ub370\uc774\ud130\ub97c \uc218\uc9d1\ud558\uace0 \ubd84\uc11d\ud558\uc5ec, \uc2e4\uc2dc\uac04\uc73c\ub85c \uc790\ub3d9\uc73c\ub85c \ub9e4\ub9e4 \uacb0\uc815\uc744 \ub0b4\ub9bd\ub2c8\ub2e4. \uc54c\uace0\ub9ac\uc998\uc758 \uc8fc\ub41c \ubaa9\uc801\uc740 \uc778\uac04\uc758 \uac10\uc815\uc774 \ubc30\uc81c\ub41c \uc0c1\ud0dc\uc5d0\uc11c \ucd5c\uc801\uc758 \uac70\ub798\ub97c \ud558\ub294 \uac83\uc785\ub2c8\ub2e4.<\/p>\n
\uba38\uc2e0\ub7ec\ub2dd(Machine Learning)\uacfc \ub525\ub7ec\ub2dd(Deep Learning)\uc740 \ub370\uc774\ud130\uc5d0\uc11c \ud328\ud134\uc744 \ud559\uc2b5\ud558\uc5ec \uc608\uce21 \uac00\ub2a5\ud55c \ubaa8\ub378\uc744 \ub9cc\ub4dc\ub294 \uae30\uc220\uc785\ub2c8\ub2e4. \uc774\ub4e4 \uae30\uc220\uc740 \ud1b5\uacc4\ud559, \ucef4\ud4e8\ud130 \uacfc\ud559, \ub370\uc774\ud130 \ubd84\uc11d\uc758 \ubb38\uc81c\ub97c \ud574\uacb0\ud558\uae30 \uc704\ud574 \uc0ac\uc6a9\ub429\ub2c8\ub2e4.<\/p>\n
\uba38\uc2e0\ub7ec\ub2dd\uc740 \ub370\uc774\ud130\ub97c \ud1b5\ud55c \ud559\uc2b5\uc744 \ud1b5\ud574 \uc608\uce21\uc744 \uc218\ud589\ud558\ub294 \uc54c\uace0\ub9ac\uc998\uc785\ub2c8\ub2e4. \uc77c\ubc18\uc801\uc73c\ub85c \ub2e4\uc74c\uacfc \uac19\uc740 \uc138 \uac00\uc9c0 \uc720\ud615\uc774 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
\ub525\ub7ec\ub2dd\uc740 \uba38\uc2e0\ub7ec\ub2dd\uc758 \ud55c \ubd84\uc57c\ub85c, \uc5ec\ub7ec \uce35\uc758 \uc778\uacf5\uc2e0\uacbd\ub9dd\uc744 \uc0ac\uc6a9\ud558\uc5ec \ub370\uc774\ud130\ub97c \ubd84\uc11d\ud558\ub294 \ubc29\uc2dd\uc785\ub2c8\ub2e4. \uc774\ubbf8\uc9c0 \uc778\uc2dd, \uc790\uc5f0\uc5b4 \ucc98\ub9ac \ubc0f \ub9ce\uc740 \ub2e4\ub978 \ub370\uc774\ud130 \uc720\ud615\uc744 \ucc98\ub9ac\ud558\ub294 \ub370 \ub9e4\uc6b0 \ud6a8\uacfc\uc801\uc785\ub2c8\ub2e4.<\/p>\n
\uc2ec\uce35 \uac15\ud654\ud559\uc2b5(Deep Reinforcement Learning, DRL)\uc740 \uac15\ud654\ud559\uc2b5\uc758 \uc6d0\ub9ac\ub97c \ubcf5\uc7a1\ud55c \uc0c1\ud0dc \uacf5\uac04\uc5d0\uc11c\ub3c4 \uc801\uc6a9\ud560 \uc218 \uc788\ub3c4\ub85d \ud574\uc8fc\ub294 \uae30\uc220\ub85c, \uc2ec\uce35 \uc2e0\uacbd\ub9dd\uc744 \uc0ac\uc6a9\ud558\uc5ec \ud658\uacbd\uc744 \ubaa8\ub378\ub9c1\ud569\ub2c8\ub2e4.<\/p>\n
OpenAI Gym\uc740 \uac15\ud654\ud559\uc2b5\uc744 \uc704\ud55c \ub2e4\uc591\ud55c \ud658\uacbd\uc744 \uc81c\uacf5\ud558\ub294 \ud234\ud0b7\uc785\ub2c8\ub2e4. \uc774\ub294 \uc5f0\uad6c\uc790 \ubc0f \uac1c\ubc1c\uc790\uac00 \uc790\uc2e0\uc758 \uc54c\uace0\ub9ac\uc998\uc744 \uc27d\uac8c \ud14c\uc2a4\ud2b8\ud558\uace0 \ube44\uad50\ud560 \uc218 \uc788\ub3c4\ub85d \ub3d5\uc2b5\ub2c8\ub2e4. Gym\uc740 \ub2e4\uc591\ud55c \ud658\uacbd\uc744 \uc81c\uacf5\ud558\uba70, \uadf8 \uc911\uc5d0\ub294 \uac8c\uc784, \ub85c\ubd07 \uc2dc\ubbac\ub808\uc774\uc158, \uae08\uc735 \uc2dc\ubbac\ub808\uc774\uc158 \ub4f1\uc774 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
\uc774\uc81c \uc2ec\uce35 \uac15\ud654\ud559\uc2b5\uc744 \uc0ac\uc6a9\ud558\uc5ec \uac04\ub2e8\ud55c \ud2b8\ub808\uc774\ub529 \uc54c\uace0\ub9ac\uc998\uc744 \uad6c\ud604\ud558\ub294 \ubc29\ubc95\uc744 \ub2e8\uacc4\ubcc4\ub85c \uc124\uba85\ud558\uaca0\uc2b5\ub2c8\ub2e4.<\/p>\n
\n# \ud544\uc694\ud55c \ub77c\uc774\ube0c\ub7ec\ub9ac \uc124\uce58\n!pip install gym numpy matplotlib\n <\/pre>\n4.2 \uae08\uc735 \uac70\ub798 \ud658\uacbd \uc0dd\uc131<\/h3>\n
\uc2e4\uc81c \uae08\uc735 \uac70\ub798\ub97c \uc2dc\ubbac\ub808\uc774\uc158\ud558\uae30 \uc704\ud574 Gym \ud658\uacbd\uc744 \uc124\uc815\ud574\uc57c \ud569\ub2c8\ub2e4. \uc774\ub97c \uc704\ud574 OpenAI Gym\uc758 Custom Environment\ub97c \ub9cc\ub4e4\uc5b4\uc57c \ud569\ub2c8\ub2e4.<\/p>\n
\nimport gym\nfrom gym import spaces\nimport numpy as np\n\nclass StockTradingEnv(gym.Env):\n def __init__(self, stock_data):\n super(StockTradingEnv, self).__init__()\n self.stock_data = stock_data\n self.current_step = 0\n self.action_space = spaces.Discrete(3) # 0: \ubcf4\uc720, 1: \ub9e4\uc218, 2: \ub9e4\ub3c4\n self.observation_space = spaces.Box(low=0, high=np.inf, shape=(len(stock_data.columns),), dtype=np.float32)\n\n def reset(self):\n self.current_step = 0\n return self.stock_data.iloc[self.current_step].values\n\n def step(self, action):\n # \ud589\ub3d9\uc5d0 \ub530\ub978 \ubcf4\uc0c1 \uacc4\uc0b0 \ubc0f \uc0c1\ud0dc \uc5c5\ub370\uc774\ud2b8\n ...\n return next_state, reward, done, {}\n <\/pre>\n4.3 \uc2ec\uce35 \uc2e0\uacbd\ub9dd \ubaa8\ub378 \uc124\uacc4<\/h3>\n
\uc8fc\uc2dd \ub9e4\ub9e4\ub97c \uc704\ud55c \ubaa8\ub378\uc744 \uc124\uacc4\ud569\ub2c8\ub2e4. \uc774\ub97c \uc704\ud574 Keras \ub610\ub294 PyTorch \uac19\uc740 \ub77c\uc774\ube0c\ub7ec\ub9ac\ub97c \uc0ac\uc6a9\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
\nfrom keras.models import Sequential\nfrom keras.layers import Dense\n\ndef create_model(input_shape):\n model = Sequential()\n model.add(Dense(24, input_shape=(input_shape,), activation='relu'))\n model.add(Dense(24, activation='relu'))\n model.add(Dense(3, activation='linear')) # \ud589\ub3d9\uc758 \uc218\uc5d0 \ub530\ub77c \ucd9c\ub825 \ub178\ub4dc\n model.compile(optimizer='adam', loss='mse')\n return model\n <\/pre>\n4.4 \ud559\uc2b5 \ub8e8\ud504 \uad6c\ud604<\/h3>\n
\ubaa8\ub378\uc744 \ud559\uc2b5\uc2dc\ud0a4\uae30 \uc704\ud55c \ub8e8\ud504\ub97c \uad6c\ud604\ud569\ub2c8\ub2e4.<\/p>\n
\nfor episode in range(num_episodes):\n state = env.reset()\n done = False\n while not done:\n # \ubaa8\ub378\uc744 \uc0ac\uc6a9\ud558\uc5ec \ud589\ub3d9 \uc120\ud0dd\n ...\n # \ud658\uacbd\uc5d0\uc11c \ub2e4\uc74c \uc0c1\ud0dc, \ubcf4\uc0c1 \uad00\ucc30\n next_state, reward, done, _ = env.step(action)\n # Q-\ub7ec\ub2dd \uc5c5\ub370\uc774\ud2b8\n ...\n state = next_state\n <\/pre>\n5. \uc131\ub2a5 \ud3c9\uac00 \ubc0f \uac15\ud654<\/h2>\n
\ubaa8\ub378\uc758 \ud559\uc2b5\uc774 \uc644\ub8cc\ub41c \ud6c4, \ud14c\uc2a4\ud2b8 \ub370\uc774\ud130\ub97c \ud1b5\ud574 \uc131\ub2a5\uc744 \ud3c9\uac00\ud569\ub2c8\ub2e4. \uc218\uc775\ub960, \ubcc0\ub3d9\uc131 \ubc0f \ucd5c\ub300 \ud558\ub77d\ub960 \uac19\uc740 \uc9c0\ud45c\ub97c \uc0ac\uc6a9\ud558\uc5ec \uc131\uacfc\ub97c \uce21\uc815\ud569\ub2c8\ub2e4. \uc774\ud6c4 \ud558\uc774\ud37c\ud30c\ub77c\ubbf8\ud130 \ud29c\ub2dd \ubc0f \ub2e4\uc591\ud55c \uae30\ubc95\uc744 \uc801\uc6a9\ud558\uc5ec \ubaa8\ub378 \uc131\ub2a5\uc744 \uac15\ud654\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
5.1 \uacb0\uacfc \uc2dc\uac01\ud654<\/h3>\n
\uc8fc\uc2dd \uac00\uaca9\uacfc \ubaa8\ub378\uc758 \ub9e4\ub9e4 \uacb0\uc815\uc744 \uc2dc\uac01\ud654\ud558\uc5ec \uacb0\uacfc\ub97c \ubd84\uc11d\ud569\ub2c8\ub2e4.<\/p>\n
\nimport matplotlib.pyplot as plt\n\nplt.plot(test_data['Close'], label='Actual Price')\nplt.plot(predicted_prices, label='Predicted Price')\nplt.legend()\nplt.show()\n <\/pre>\n\uacb0\ub860<\/h2>\n
\uc2ec\uce35 \uac15\ud654\ud559\uc2b5\uc740 \uc54c\uace0\ub9ac\uc998 \ud2b8\ub808\uc774\ub529\uc758 \ubbf8\ub798\ub97c \uc5f4\uc5b4\uc8fc\ub294 \ud601\uc2e0\uc801\uc778 \uae30\uc220\uc785\ub2c8\ub2e4. OpenAI Gym\uc744 \ud1b5\ud574 \uac15\ud654\ud559\uc2b5\uc744 \uc2e4\ud5d8\ud558\uace0, \ub2e4\uc591\ud55c \uae08\uc735 \ud658\uacbd\uc5d0\uc11c \ud2b8\ub808\uc774\ub529 \ubaa8\ub378\uc744 \uc0dd\uc131\ud560 \uc218 \uc788\ub294 \uac00\ub2a5\uc131\uc774 \ubb34\uad81\ubb34\uc9c4\ud569\ub2c8\ub2e4. \uc774 \uac15\uc88c\ub97c \ud1b5\ud574 \ubc30\uc6b4 \ub0b4\uc6a9\uc744 \ubc14\ud0d5\uc73c\ub85c \uc790\uc2e0\ub9cc\uc758 \ud2b8\ub808\uc774\ub529 \uc54c\uace0\ub9ac\uc998\uc744 \ub9cc\ub4dc\uc2dc\uae38 \ubc14\ub78d\ub2c8\ub2e4.<\/p>\n
\ucc38\uace0 \uc790\ub8cc<\/h2>\n