{"id":4507,"date":"2023-07-10T13:36:46","date_gmt":"2023-07-10T13:36:46","guid":{"rendered":"https:\/\/statorials.org\/tr\/tren-kontrolu-r\/"},"modified":"2023-07-10T13:36:46","modified_gmt":"2023-07-10T13:36:46","slug":"tren-kontrolu-r","status":"publish","type":"post","link":"https:\/\/statorials.org\/tr\/tren-kontrolu-r\/","title":{"rendered":"C: e\u011fitim parametrelerini kontrol etmek i\u00e7in traincontrol nas\u0131l kullan\u0131l\u0131r?"},"content":{"rendered":"

<\/p>\n

\n

Bir modelin bir veri k\u00fcmesine ne kadar iyi uydu\u011funu de\u011ferlendirmek i\u00e7in, daha \u00f6nce hi\u00e7 g\u00f6rmedi\u011fi g\u00f6zlemlerdeki performans\u0131n\u0131 analiz etmemiz gerekir.<\/span><\/p>\n

Bunu ba\u015farman\u0131n en yayg\u0131n yollar\u0131ndan biri, a\u015fa\u011f\u0131daki yakla\u015f\u0131m\u0131 kullanan k-katl\u0131 \u00e7apraz do\u011frulamay\u0131<\/a> kullanmakt\u0131r:<\/span><\/p>\n

1.<\/strong> Bir veri k\u00fcmesini rastgele olarak yakla\u015f\u0131k olarak e\u015fit b\u00fcy\u00fckl\u00fckte k<\/em> gruba veya “katlamaya” b\u00f6l\u00fcn.<\/span><\/p>\n

2.<\/strong> Tutucu set olarak k\u0131vr\u0131mlardan birini se\u00e7in. \u015eablonu kalan k-1 k\u0131vr\u0131mlar\u0131na g\u00f6re ayarlay\u0131n. Gerilmi\u015f kattaki g\u00f6zlemler \u00fczerinden MSE testini hesaplay\u0131n.<\/span><\/p>\n

3.<\/strong> Her seferinde farkl\u0131 bir k\u00fcmeyi d\u0131\u015flama k\u00fcmesi olarak kullanarak bu i\u015flemi k<\/em> kez tekrarlay\u0131n.<\/span><\/p>\n

4.<\/strong> Genel test MSE’sini k<\/em> test MSE’sinin ortalamas\u0131 olarak hesaplay\u0131n.<\/span><\/p>\n

R’de k-katl\u0131 \u00e7apraz do\u011frulama ger\u00e7ekle\u015ftirmenin en kolay yolu, R’deki \u015fapka<\/strong> k\u00fct\u00fcphanesindeki trainControl()<\/strong> ve train()<\/strong> i\u015flevlerini kullanmakt\u0131r.<\/span><\/p>\n

trainControl()<\/strong> i\u015flevi, e\u011fitim parametrelerini belirlemek i\u00e7in kullan\u0131l\u0131r (\u00f6rne\u011fin, kullan\u0131lacak \u00e7apraz do\u011frulama t\u00fcr\u00fc, kullan\u0131lacak katlama say\u0131s\u0131, vb.) ve train()<\/strong> i\u015flevi, modeli ger\u00e7ekten verilere s\u0131\u011fd\u0131rmak i\u00e7in kullan\u0131l\u0131r. .<\/span><\/p>\n

A\u015fa\u011f\u0131daki \u00f6rnek, trainControl()<\/strong> ve train()<\/strong> i\u015flevlerinin pratikte nas\u0131l kullan\u0131laca\u011f\u0131n\u0131 g\u00f6sterir.<\/span><\/p>\n

\u00d6rnek: R’de trainControl() nas\u0131l kullan\u0131l\u0131r?<\/strong><\/span><\/h2>\n

R’de a\u015fa\u011f\u0131daki veri k\u00fcmesine sahip oldu\u011fumuzu varsayal\u0131m:<\/span><\/p>\n

 #create data frame<\/span>\ndf <- data.frame(y=c(6, 8, 12, 14, 14, 15, 17, 22, 24, 23),\n                 x1=c(2, 5, 4, 3, 4, 6, 7, 5, 8, 9),\n                 x2=c(14, 12, 12, 13, 7, 8, 7, 4, 6, 5))\n\n#view data frame\n<\/span>df\n\ny x1 x2\n6 2 14\n8 5 12\n12 4 12\n14 3 13\n14 4 7\n15 6 8\n17 7 7\n22 5 4\n24 8 6\n23 9 5\n<\/strong><\/pre>\n
 \u015eimdi bu veri setine \u00e7oklu do\u011frusal regresyon modelini<\/a> s\u0131\u011fd\u0131rmak i\u00e7in lm()<\/a> fonksiyonunu kulland\u0131\u011f\u0131m\u0131z\u0131 varsayal\u0131m; tahmin de\u011fi\u015fkenleri olarak x1<\/strong> ve x2’yi<\/strong> ve yan\u0131t de\u011fi\u015fkeni olarak y’yi<\/strong> kullan\u0131yoruz:<\/span><\/p>\n
 #fit multiple linear regression model to data<\/span>\nfit <- lm(y ~ x1 + x2, data=df)\n\n#view model summary\n<\/span>summary(fit)\n\nCall:\nlm(formula = y ~ x1 + x2, data = df)\n\nResiduals:\n    Min 1Q Median 3Q Max \n-3.6650 -1.9228 -0.3684 1.2783 5.0208 \n\nCoefficients:\n            Estimate Std. Error t value Pr(>|t|)  \n(Intercept) 21.2672 6.9927 3.041 0.0188 *\nx1 0.7803 0.6942 1.124 0.2981  \nx2 -1.1253 0.4251 -2.647 0.0331 *\n---\nSignificant. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1\n\nResidual standard error: 3.093 on 7 degrees of freedom\nMultiple R-squared: 0.801, Adjusted R-squared: 0.7441 \nF-statistic: 14.09 on 2 and 7 DF, p-value: 0.003516\n<\/strong><\/pre>\n
 Model \u00e7\u0131kt\u0131s\u0131ndaki katsay\u0131lar\u0131 kullanarak uygun regresyon modelini yazabiliriz:<\/span><\/p>\n
 y = 21,2672 + 0,7803*(x 1<\/sub> ) \u2013 1,1253(x 2<\/sub> )<\/strong><\/span><\/p>\n
 Bu modelin g\u00f6r\u00fcnmeyen g\u00f6zlemlerde<\/a> ne kadar iyi performans g\u00f6sterdi\u011fine dair bir fikir edinmek i\u00e7in k-katl\u0131 \u00e7apraz do\u011frulamay\u0131 kullanabiliriz.<\/span><\/p>\n
 A\u015fa\u011f\u0131daki kod, 5 katlama ( number=5<\/strong> ) kullanan k katl\u0131 \u00e7apraz do\u011frulamay\u0131 ( method=”cv”<\/strong> ) belirtmek i\u00e7in d\u00fczeltme i\u015fareti<\/strong> paketinin trainControl()<\/strong> i\u015flevinin nas\u0131l kullan\u0131laca\u011f\u0131n\u0131 g\u00f6sterir.<\/span><\/p>\n
 Daha sonra k-katl\u0131 \u00e7apraz do\u011frulamay\u0131 ger\u00e7ekle\u015ftirmek i\u00e7in bu trainControl()<\/strong> fonksiyonunu train()<\/strong> fonksiyonuna aktar\u0131r\u0131z:<\/span><\/p>\n
 library<\/span> (caret)<\/span>\n\n#specify the cross-validation method<\/span>\nctrl <- trainControl(method = \" cv<\/span> \", number = 5<\/span> )\n\n#fit a regression model and use k-fold CV to evaluate performance\n<\/span>model <- train(y ~ x1 + x2, data = df, method = \" lm<\/span> \", trControl = ctrl)\n\n#view summary of k-fold CV               \n<\/span>print<\/span> (model)\n\nLinear Regression \n\n10 samples\n 2 predictors\n\nNo pre-processing\nResampling: Cross-Validated (5 fold) \nSummary of sample sizes: 8, 8, 8, 8, 8 \nResampling results:\n\n  RMSE Rsquared MAE     \n  3.612302 1 3.232153\n\nTuning parameter 'intercept' was held constant at a value of TRUE\n<\/strong><\/pre>\n
 Sonu\u00e7tan, her seferinde 8<\/strong> g\u00f6zlemlik bir \u00f6rneklem b\u00fcy\u00fckl\u00fc\u011f\u00fc kullan\u0131larak modelin 5<\/strong> kez uyduruldu\u011funu g\u00f6rebiliriz.<\/span><\/p>\n
 Her seferinde model, tutulan 2<\/strong> g\u00f6zlemin de\u011ferlerini tahmin etmek i\u00e7in kullan\u0131ld\u0131 ve her seferinde a\u015fa\u011f\u0131daki \u00f6l\u00e7\u00fcmler hesapland\u0131:<\/span><\/p>\n