\u00c7o\u011fu zaman, birden fazla s\u0131n\u0131fland\u0131rma modelini tek bir veri k\u00fcmesine s\u0131\u011fd\u0131rmak ve hangi modelin verilerde en iyi performans\u0131 g\u00f6sterdi\u011fini g\u00f6rselle\u015ftirmek i\u00e7in her model i\u00e7in bir ROC e\u011frisi olu\u015fturmak isteyebilirsiniz.<\/span><\/p>\n A\u015fa\u011f\u0131daki ad\u0131m ad\u0131m \u00f6rnek, Python’da birden fazla ROC e\u011frisinin nas\u0131l \u00e7izilece\u011fini g\u00f6sterir.<\/span><\/p>\n \u00d6ncelikle gerekli birka\u00e7 paketi Python’a aktaraca\u011f\u0131z:<\/span><\/p>\n Daha sonra, 1000 sat\u0131r, d\u00f6rt \u00f6ng\u00f6r\u00fcc\u00fc de\u011fi\u015fken ve bir ikili yan\u0131t de\u011fi\u015fkeni i\u00e7eren sahte bir veri k\u00fcmesi olu\u015fturmak i\u00e7in sklearn’in make_classification()<\/a> i\u015flevini kullanaca\u011f\u0131z:<\/span><\/p>\n Daha sonra, verilere bir lojistik regresyon modeli ve ard\u0131ndan gradyanla geli\u015ftirilmi\u015f bir model yerle\u015ftirece\u011fiz ve her model i\u00e7in ROC e\u011frisini ayn\u0131 \u00e7izim \u00fczerinde \u00e7izece\u011fiz:<\/span> <\/p>\n Mavi \u00e7izgi lojistik regresyon modeli i\u00e7in ROC e\u011frisini, turuncu \u00e7izgi ise gradyan destekli model i\u00e7in ROC e\u011frisini g\u00f6sterir.<\/span><\/p>\n ROC e\u011frisi grafi\u011fin sol \u00fcst k\u00f6\u015fesine ne kadar yak\u0131n olursa, model verileri kategorilere g\u00f6re o kadar iyi s\u0131n\u0131fland\u0131rabilir.<\/span><\/p>\n Bunu \u00f6l\u00e7mek i\u00e7in, bize grafi\u011fin ne kadar\u0131n\u0131n e\u011frinin alt\u0131nda oldu\u011funu s\u00f6yleyen AUC’yi (e\u011frinin alt\u0131ndaki alan\u0131) hesaplayabiliriz.<\/span><\/p>\n AUC 1’e ne kadar yak\u0131nsa model o kadar iyidir.<\/span><\/p>\n Grafi\u011fimizde her model i\u00e7in a\u015fa\u011f\u0131daki AUC \u00f6l\u00e7\u00fcmlerini g\u00f6rebiliriz:<\/span><\/p>\n A\u00e7\u0131k\u00e7as\u0131, gradyanla geli\u015ftirilmi\u015f model, verileri kategorilere ay\u0131rmada lojistik regresyon modelinden daha ba\u015far\u0131l\u0131d\u0131r.<\/span><\/p>\n A\u015fa\u011f\u0131daki e\u011fitimler s\u0131n\u0131fland\u0131rma modelleri ve ROC e\u011frileri hakk\u0131nda ek bilgi sa\u011flar:<\/span><\/p>\n Lojistik Regresyona Giri\u015f<\/a> Makine \u00f6\u011freniminde s\u0131n\u0131fland\u0131rma modellerinin performans\u0131n\u0131 g\u00f6rselle\u015ftirmenin bir yolu, “al\u0131c\u0131 \u00e7al\u0131\u015fma karakteristi\u011fi” e\u011frisi anlam\u0131na gelen bir ROC e\u011frisi olu\u015fturmakt\u0131r. \u00c7o\u011fu zaman, birden fazla s\u0131n\u0131fland\u0131rma modelini tek bir veri k\u00fcmesine s\u0131\u011fd\u0131rmak ve hangi modelin verilerde en iyi performans\u0131 g\u00f6sterdi\u011fini g\u00f6rselle\u015ftirmek i\u00e7in her model i\u00e7in bir ROC e\u011frisi olu\u015fturmak isteyebilirsiniz. A\u015fa\u011f\u0131daki ad\u0131m ad\u0131m \u00f6rnek, Python’da birden fazla ROC […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[11],"tags":[],"class_list":["post-3029","post","type-post","status-publish","format-standard","hentry","category-rehber"],"yoast_head":"\n Ad\u0131m 1: Gerekli paketleri i\u00e7e aktar\u0131n<\/strong><\/span><\/h3>\n
from<\/span> sklearn import<\/span> metrics\nfrom<\/span> sklearn import<\/span> datasets\nfrom<\/span> sklearn. model_selection<\/span> import<\/span> train_test_split\nfrom<\/span> sklearn. linear_model<\/span> import<\/span> LogisticRegression\nfrom<\/span> sklearn. set<\/span> import<\/span> GradientBoostingClassifier\nimport<\/span> numpy as<\/span> np\nimport<\/span> matplotlib. pyplot<\/span> as<\/span> plt\n<\/strong><\/span><\/pre>\n 2. Ad\u0131m: Sahte Veri Olu\u015fturun<\/strong><\/span><\/h3>\n
#create fake dataset\n<\/span>X, y = datasets. make_classification<\/span> (n_samples= 1000<\/span> ,\n n_features= 4<\/span> ,\n n_informative= 3<\/span> ,\n n_redundant= 1<\/span> ,\n random_state= 0<\/span> )\n\n#split dataset into training and testing set\n<\/span>X_train, X_test, y_train, y_test = train_test_split(X, y, test_size= .3<\/span> , random_state= 0<\/span> )<\/strong><\/span><\/pre>\n Ad\u0131m 3: Birden Fazla Modeli S\u0131\u011fd\u0131r\u0131n ve ROC E\u011frilerini \u00c7izin<\/strong><\/span><\/h3>\n
#set up plotting area\n<\/span>plt. Figure<\/span> (0). clf<\/span> ()\n\n#fit logistic regression model and plot ROC curve\n<\/span>model = LogisticRegression()\nmodel. fit<\/span> (X_train, y_train)\ny_pred = model. predict_proba<\/span> (X_test)[:, 1]\nfpr, tpr, _ = metrics. roc_curve<\/span> (y_test, y_pred)\nauc = round(metrics. roc_auc_score<\/span> (y_test, y_pred), 4)\nplt. plot<\/span> (fpr,tpr,label=\"Logistic Regression, AUC=\"+str(auc))\n\n#fit gradient boosted model and plot ROC curve\n<\/span>model = GradientBoostingClassifier()\nmodel. fit<\/span> (X_train, y_train)\ny_pred = model. predict_proba<\/span> (X_test)[:, 1]\nfpr, tpr, _ = metrics. roc_curve<\/span> (y_test, y_pred)\nauc = round(metrics. roc_auc_score<\/span> (y_test, y_pred), 4)\nplt. plot<\/span> (fpr,tpr,label=\"Gradient Boosting, AUC=\"+str(auc))\n\n#add legend\n<\/span>plt. legend<\/span> ()<\/strong><\/span> <\/pre>\n![\"Python'da](\"https:\/\/statorials.org\/wp-content\/uploads\/2023\/08\/multipleroc1.jpg\")
<\/p>\n\n
Ek kaynaklar<\/strong><\/span><\/h3>\n
ROC E\u011frisi Nas\u0131l Yorumlan\u0131r (\u00d6rneklerle)<\/a>
\u0130yi bir AUC puan\u0131 nedir?<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"