\ud000\ud2b8 \ub9e4\ub9e4, \uc989 \uc54c\uace0\ub9ac\uc998 \ud2b8\ub808\uc774\ub529\uc740 \ud604\ub300 \uae08\uc735 \uc2dc\uc7a5\uc758 \uc911\uc694\ud55c \ubd80\ubd84\uc785\ub2c8\ub2e4. \uba38\uc2e0\ub7ec\ub2dd(Machine Learning)\uacfc \ub525\ub7ec\ub2dd(Deep Learning) \uae30\uc220\uc774 \ubc1c\uc804\ud568\uc5d0 \ub530\ub77c \uc774\ub7ec\ud55c \uae30\uc220\ub4e4\uc744 \ud65c\uc6a9\ud55c \uc790\ub3d9 \ub9e4\ub9e4 \uc2dc\uc2a4\ud15c\uc774 \ub9ce\uc740 \uc8fc\ubaa9\uc744 \ubc1b\uace0 \uc788\uc2b5\ub2c8\ub2e4. \ubcf8 \uae00\uc5d0\uc11c\ub294 XGBoost, LightGBM, CatBoost\uc640 \uac19\uc740 \uac15\ub825\ud55c \uba38\uc2e0\ub7ec\ub2dd \uc54c\uace0\ub9ac\uc998\uc744 \ud65c\uc6a9\ud558\uc5ec \ud6a8\uacfc\uc801\uc778 \ud2b8\ub808\uc774\ub529 \uc804\ub7b5\uc744 \uad6c\ud604\ud558\ub294 \ubc29\ubc95\uc5d0 \ub300\ud574 \ub2e4\ub8f9\ub2c8\ub2e4.<\/p>\n
\uba38\uc2e0\ub7ec\ub2dd\uc740 \ub370\uc774\ud130\ub97c \ubd84\uc11d\ud558\uace0, \uc774 \ub370\uc774\ud130\ub85c\ubd80\ud130 \ud559\uc2b5\ud558\uc5ec \uc608\uce21\uc774\ub098 \uacb0\uc815\uc744 \ub0b4\ub9ac\ub294 \uc54c\uace0\ub9ac\uc998\uc744 \uac1c\ubc1c\ud558\ub294 \ubd84\uc57c\uc785\ub2c8\ub2e4. \uc77c\ubc18\uc801\uc73c\ub85c \uba38\uc2e0\ub7ec\ub2dd\uc740 \uc9c0\ub3c4\ud559\uc2b5, \ube44\uc9c0\ub3c4\ud559\uc2b5, \uc900\uc9c0\ub3c4\ud559\uc2b5 \ub4f1\uc73c\ub85c \ubd84\ub958\ub429\ub2c8\ub2e4. \uc774\ub97c \ud1b5\ud574 \uc6b0\ub9ac\ub294 \ub2e4\uc591\ud55c \uc720\ud615\uc758 \ub370\uc774\ud130\uc5d0\uc11c \ud328\ud134\uc744 \ucc3e\uc544\ub0bc \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
\ub525\ub7ec\ub2dd\uc740 \uba38\uc2e0\ub7ec\ub2dd\uc758 \ud558\uc704 \ubd84\uc57c\ub85c, \uc778\uacf5\uc2e0\uacbd\ub9dd\uc744 \ud65c\uc6a9\ud558\uc5ec \ub370\uc774\ud130\ub97c \ud559\uc2b5\ud569\ub2c8\ub2e4. \ud2b9\ud788 \ubcf5\uc7a1\ud55c \ub370\uc774\ud130 \uad6c\uc870, \uc608\ub97c \ub4e4\uc5b4 \uc774\ubbf8\uc9c0\ub098 \uc790\uc5f0\uc5b4 \ucc98\ub9ac\ub97c \ucc98\ub9ac\ud558\ub294 \ub370 \ub6f0\uc5b4\ub09c \uc131\ub2a5\uc744 \ubc1c\ud718\ud569\ub2c8\ub2e4. \ub525\ub7ec\ub2dd \ubaa8\ub378\uc740 \uc77c\ubc18\uc801\uc73c\ub85c \ub2e4\uce35 \uc2e0\uacbd\ub9dd\uc73c\ub85c \uad6c\uc131\ub418\uc5b4 \uc788\uc73c\uba70, \ub354 \ub9ce\uc740 \ub9e4\uac1c\ubcc0\uc218\uc640 \ubcf5\uc7a1\uc131\uc744 \uac00\uc9c4 \uacbd\uc6b0\uac00 \ub9ce\uc2b5\ub2c8\ub2e4.<\/p>\n
\uc54c\uace0\ub9ac\uc998 \ud2b8\ub808\uc774\ub529\uc5d0\uc11c \uba38\uc2e0\ub7ec\ub2dd\uc758 \uac00\uc7a5 \ud070 \uc7a5\uc810\uc740 \ub370\uc774\ud130 \uae30\ubc18\uc758 \uacb0\uc815 \ud504\ub85c\uc138\uc2a4\ub97c \uc790\ub3d9\ud654\ud558\uc5ec \uc778\uac04\uc758 \uac10\uc815\uc774\ub098 \ud3b8\ud5a5\uc744 \ubc30\uc81c\ud560 \uc218 \uc788\ub2e4\ub294 \uac83\uc785\ub2c8\ub2e4. \ub610\ud55c, \ubaa8\ub378\uc740 \uc2dc\uc7a5 \ub370\uc774\ud130\ub97c \ud1b5\ud574 \uc9c0\uc18d\uc801\uc73c\ub85c \ud559\uc2b5\ud558\uace0 \uac1c\uc120\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
XGBoost(eXtreme Gradient Boosting)\ub294 Gradient Boosting \uc54c\uace0\ub9ac\uc998\uc744 \uae30\ubc18\uc73c\ub85c \ud55c \uac15\ub825\ud55c \uba38\uc2e0\ub7ec\ub2dd \ub77c\uc774\ube0c\ub7ec\ub9ac\uc785\ub2c8\ub2e4. \ud2b9\ud788 \ub192\uc740 \uc608\uce21 \uc131\ub2a5\uacfc \uc18d\ub3c4\ub85c \uc778\ud574 \ub370\uc774\ud130 \uacfc\ud559 \ubc0f \uba38\uc2e0\ub7ec\ub2dd \ub300\ud68c\uc5d0\uc11c \uc790\uc8fc \uc0ac\uc6a9\ub429\ub2c8\ub2e4.<\/p>\n
\nimport xgboost as xgb\n\nmodel = xgb.XGBClassifier()\nmodel.fit(X_train, y_train)\npredictions = model.predict(X_test)\n <\/code><\/pre>\nLightGBM<\/h2>\nLightGBM\uc774\ub780?<\/h3>\n
LightGBM\uc740 Microsoft\uc5d0\uc11c \uac1c\ubc1c\ud55c Gradient Boosting \ud504\ub808\uc784\uc6cc\ud06c\ub85c, \ud2b9\ud788 \ub300\uaddc\ubaa8 \ub370\uc774\ud130\uc14b\uc5d0\uc11c \ud6a8\uc728\uc801\uc778 \uc131\ub2a5\uc744 \ubc1c\ud718\ud569\ub2c8\ub2e4. LightGBM\uc740 Histogram \uae30\ubc18\uc758 \uc54c\uace0\ub9ac\uc998\uc744 \uc0ac\uc6a9\ud558\uc5ec \ud559\uc2b5 \uc18d\ub3c4\ub97c \ud06c\uac8c \ud5a5\uc0c1\uc2dc\ud0b5\ub2c8\ub2e4.<\/p>\n
LightGBM\uc758 \uc7a5\uc810<\/h3>\n\n- \ub192\uc740 \uc131\ub2a5: \ub300\uaddc\ubaa8 \ub370\uc774\ud130\uc14b\uc5d0\uc11c\ub3c4 \uc88b\uc740 \uc131\ub2a5\uc744 \uc720\uc9c0\ud569\ub2c8\ub2e4.<\/li>\n
- \ube60\ub978 \ud6c8\ub828: Histogram \uae30\ubc18\uc758 \uc54c\uace0\ub9ac\uc998\uc73c\ub85c \ube60\ub978 \ud6c8\ub828\uc744 \uc9c0\uc6d0\ud569\ub2c8\ub2e4.<\/li>\n
- \uba54\ubaa8\ub9ac \ud6a8\uc728\uc131: \uba54\ubaa8\ub9ac \uc0ac\uc6a9\uc744 \ucd5c\uc18c\ud654\ud558\uc5ec \ub354 \ub9ce\uc740 \ub370\uc774\ud130\ub97c \ucc98\ub9ac\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/li>\n<\/ul>\n
LightGBM \uc0ac\uc6a9\ubc95<\/h3>\n\nimport lightgbm as lgb\n\nmodel = lgb.LGBMClassifier()\nmodel.fit(X_train, y_train)\npredictions = model.predict(X_test)\n <\/code><\/pre>\nCatBoost<\/h2>\nCatBoost\ub780?<\/h3>\n
CatBoost\ub294 Yandex\uc5d0\uc11c \uac1c\ubc1c\ud55c Gradient Boosting \ub77c\uc774\ube0c\ub7ec\ub9ac\ub85c, \ubc94\uc8fc\ud615 \ubcc0\uc218 \ucc98\ub9ac\uc5d0 \ud2b9\ud654\ub418\uc5b4 \uc788\uc2b5\ub2c8\ub2e4. \ubc94\uc8fc\ud615 \ubcc0\uc218\uc5d0 \ub300\ud574 \uc0ac\uc804 \uc5d4\uc9c0\ub2c8\uc5b4\ub9c1 \uc5c6\uc774\ub3c4 \ub192\uc740 \uc131\ub2a5\uc744 \ub0bc \uc218 \uc788\ub294 \uc810\uc774 \ud2b9\uc9d5\uc785\ub2c8\ub2e4.<\/p>\n
CatBoost\uc758 \uc7a5\uc810<\/h3>\n\n- \uc790\ub3d9 \ubc94\uc8fc\ud615 \ubcc0\uc218 \ucc98\ub9ac: \ubcc4\ub3c4\uc758 \ub370\uc774\ud130 \ubcc0\ud658 \uc5c6\uc774\ub3c4 \ubc94\uc8fc\ud615 \ubcc0\uc218\ub97c \uc0ac\uc6a9\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/li>\n
- \uc124\uba85 \uac00\ub2a5\ud55c \ubaa8\ub378: \uc911\uc694 \ubcc0\uc218\ub97c \uc2dc\uac01\ud654\ud558\uc5ec \ubaa8\ub378\uc758 \uacb0\uacfc\ub97c \uc774\ud574\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/li>\n
- \ube60\ub978 \ud559\uc2b5 \uc18d\ub3c4: \uc18c\ud615 \ubc0f \uc911\ud615 \ub370\uc774\ud130\uc14b\uc5d0\uc11c \ube60\ub978 \ud559\uc2b5 \uc18d\ub3c4\ub97c \uc81c\uacf5\ud569\ub2c8\ub2e4.<\/li>\n<\/ul>\n
CatBoost \uc0ac\uc6a9\ubc95<\/h3>\n\nimport catboost\n\nmodel = catboost.CatBoostClassifier()\nmodel.fit(X_train, y_train)\npredictions = model.predict(X_test)\n <\/code><\/pre>\n\ubaa8\ub378 \ud559\uc2b5 \ubc0f \ud3c9\uac00<\/h2>\n
\ubaa8\ub378\uc744 \ud6c8\ub828\ud558\uace0 \ud3c9\uac00\ud558\ub294 \uacfc\uc815\uc740 \uc54c\uace0\ub9ac\uc998 \ud2b8\ub808\uc774\ub529\uc758 \uc131\ud328\ub97c \uac00\ub974\ub294 \uc911\uc694\ud55c \ub2e8\uacc4\uc785\ub2c8\ub2e4. \uc774\ub97c \uc704\ud574 \ud6c8\ub828 \ub370\uc774\ud130\uc640 \ud14c\uc2a4\ud2b8 \ub370\uc774\ud130\ub97c \ub098\ub204\uace0, \ub2e4\uc591\ud55c \uc131\ub2a5 \uc9c0\ud45c\ub97c \ubc14\ud0d5\uc73c\ub85c \ubaa8\ub378\uc744 \ud3c9\uac00\ud574\uc57c \ud569\ub2c8\ub2e4.<\/p>\n
\ud6c8\ub828 \ubc0f \ud14c\uc2a4\ud2b8 \ub370\uc774\ud130 \ubd84\ub9ac<\/h3>\n\nfrom sklearn.model_selection import train_test_split\n\nX_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)\n <\/code><\/pre>\n\ubaa8\ub378 \ud3c9\uac00 \uc9c0\ud45c<\/h3>\n
\ubaa8\ub378\uc758 \uc131\ub2a5\uc744 \ud3c9\uac00\ud558\ub294 \uc9c0\ud45c\ub85c\ub294 \uc815\ud655\ub3c4(Accuracy), \uc815\ubc00\ub3c4(Precision), \uc7ac\ud604\uc728(Recall), F1 Score \ub4f1\uc774 \uc788\uc2b5\ub2c8\ub2e4. \uc774\ub4e4 \uc9c0\ud45c\ub97c \ud65c\uc6a9\ud558\uc5ec \ubaa8\ub378\uc758 \uc131\ub2a5\uc744 \uc885\ud569\uc801\uc73c\ub85c \ud3c9\uac00\ud574\uc57c \ud569\ub2c8\ub2e4.<\/p>\n
\uacb0\ub860<\/h2>\n
XGBoost, LightGBM, CatBoost \ub4f1 \ub2e4\uc591\ud55c \uba38\uc2e0\ub7ec\ub2dd \uc54c\uace0\ub9ac\uc998\uc744 \ud65c\uc6a9\ud558\uc5ec \uc54c\uace0\ub9ac\uc998 \ud2b8\ub808\uc774\ub529 \uc2dc\uc2a4\ud15c\uc744 \uad6c\ud604\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uac01 \uc54c\uace0\ub9ac\uc998\uc758 \ud2b9\uc9d5\uacfc \uc7a5\uc810\uc744 \uc798 \uc774\ud574\ud558\uace0, \uc801\uc808\ud788 \uc801\uc6a9\ud55c\ub2e4\uba74 \ud6a8\uacfc\uc801\uc778 \uc790\ub3d9 \ub9e4\ub9e4 \uc2dc\uc2a4\ud15c\uc744 \uad6c\ucd95\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uc774\ub7ec\ud55c \uc2dc\uc2a4\ud15c\uc740 \ub370\uc774\ud130 \uae30\ubc18\uc758 \uc804\ub7b5\uc744 \ud1b5\ud574 \uc2dc\uc7a5\uc758 \ubcc0\ub3d9\uc131\uc744 \ud6a8\uacfc\uc801\uc73c\ub85c \ud65c\uc6a9\ud560 \uc218 \uc788\uac8c \ud574\uc90d\ub2c8\ub2e4.<\/p>\n
\ucc38\uace0 \uc790\ub8cc<\/h2>\n