<\/p>\n
Akaike Bilgi Kriteri ( AIC<\/strong> ), bir modelin bir veri k\u00fcmesine ne kadar iyi uydu\u011funu \u00f6l\u00e7mek i\u00e7in kullan\u0131lan bir \u00f6l\u00e7\u00fcmd\u00fcr.<\/span><\/p>\n A\u015fa\u011f\u0131daki \u015fekilde hesaplan\u0131r:<\/span><\/p>\n AIC = 2K \u2013 2 ln<\/em> (L)<\/span><\/strong><\/p>\n Alt\u0131n:<\/span><\/p>\n AIC, verilerdeki en fazla varyasyonu a\u00e7\u0131klayan modeli bulmak ve a\u015f\u0131r\u0131 say\u0131da parametre kullanan modelleri cezaland\u0131rmak i\u00e7in tasarlanm\u0131\u015ft\u0131r.<\/span><\/p>\n En d\u00fc\u015f\u00fck AIC de\u011ferine sahip modeli \u00fcreten tahmin de\u011fi\u015fkenleri (veya “\u00f6zellikler”) k\u00fcmesini bulana kadar, bir regresyon modeline tahmin de\u011fi\u015fkenlerini yinelemeli olarak eklemek ve modelden \u00e7\u0131karmak i\u00e7in R’deki MASS<\/strong> paketindeki stepAIC()<\/strong> i\u015flevini kullanabilirsiniz.<\/span><\/p>\n Bu i\u015flev a\u015fa\u011f\u0131daki temel s\u00f6zdizimini kullan\u0131r:<\/span><\/p>\n stepAIC(nesne, y\u00f6n, \u2026)<\/strong><\/span><\/p>\n Alt\u0131n:<\/span><\/p>\n A\u015fa\u011f\u0131daki \u00f6rnekte bu fonksiyonun pratikte nas\u0131l kullan\u0131laca\u011f\u0131 g\u00f6sterilmektedir.<\/span><\/p>\n Bu \u00f6rnek i\u00e7in, R’de yerle\u015fik olarak bulunan ve 32 farkl\u0131 araba i\u00e7in 11 farkl\u0131 \u00f6znitelik \u00fczerinde \u00f6l\u00e7\u00fcmler i\u00e7eren mtcars<\/a> veri k\u00fcmesini kullanaca\u011f\u0131z:<\/span><\/p>\n Yan\u0131t de\u011fi\u015fkeni olarak hp’yi<\/strong> ve a\u015fa\u011f\u0131daki potansiyel tahmin de\u011fi\u015fkenlerini kullanarak bir regresyon modeli uydurmak istedi\u011fimizi varsayal\u0131m:<\/span><\/p>\n M\u00fcmk\u00fcn olan en d\u00fc\u015f\u00fck AIC de\u011ferine sahip modele ula\u015fana kadar \u00e7e\u015fitli tahmin de\u011fi\u015fkenlerini modele eklemek ve \u00e7\u0131karmak i\u00e7in MASS<\/b> paketindeki stepAIC()<\/strong> i\u015flevini kullanabiliriz:<\/span><\/p>\n Sonucun nas\u0131l yorumlanaca\u011f\u0131 a\u015fa\u011f\u0131da a\u00e7\u0131klanm\u0131\u015ft\u0131r:<\/span><\/p>\n (1)<\/strong> \u0130lk olarak, d\u00f6rt yorday\u0131c\u0131 de\u011fi\u015fkenle bir regresyon modeli uydurarak ba\u015fl\u0131yoruz. Bu modelin AIC de\u011feri 226,88’dir<\/strong> .<\/span><\/p>\n (2)<\/strong> Daha sonra stepAIC, tahmin de\u011fi\u015fkeni olarak drat’\u0131n<\/strong> kald\u0131r\u0131lmas\u0131n\u0131n AIC de\u011ferini 224,98’e<\/strong> daha da d\u00fc\u015f\u00fcrece\u011fini belirler.<\/span><\/p>\n (3)<\/strong> Daha sonra stepAIC modeli, mpg’yi<\/strong> \u00f6ng\u00f6r\u00fcc\u00fc de\u011fi\u015fken olarak kald\u0131rman\u0131n AIC de\u011ferini 224,56’ya<\/strong> daha da d\u00fc\u015f\u00fcrece\u011fini belirler.<\/span><\/p>\n (4)<\/strong> Son olarak stepAIC, de\u011fi\u015fkenleri ekleyerek veya \u00e7\u0131kararak AIC de\u011ferini daha da azaltman\u0131n bir yolu olmad\u0131\u011f\u0131n\u0131 belirler.<\/span><\/p>\n Bu nedenle nihai model \u015fu \u015fekildedir:<\/span><\/p>\n hp = 441,26 + 38,67 (a\u011f\u0131rl\u0131k) \u2013 23,47 (qsec)<\/strong><\/span><\/p>\n Bu modelin AIC de\u011feri 224,56’d\u0131r<\/strong> .<\/span><\/p>\n A\u015fa\u011f\u0131daki e\u011fitimlerde R’de di\u011fer ortak g\u00f6revlerin nas\u0131l ger\u00e7ekle\u015ftirilece\u011fi a\u00e7\u0131klanmaktad\u0131r:<\/span><\/p>\n R’de \u00e7oklu do\u011frusal regresyon nas\u0131l ger\u00e7ekle\u015ftirilir<\/a> Akaike Bilgi Kriteri ( AIC ), bir modelin bir veri k\u00fcmesine ne kadar iyi uydu\u011funu \u00f6l\u00e7mek i\u00e7in kullan\u0131lan bir \u00f6l\u00e7\u00fcmd\u00fcr. A\u015fa\u011f\u0131daki \u015fekilde hesaplan\u0131r: AIC = 2K \u2013 2 ln (L) Alt\u0131n: K: Model parametrelerinin say\u0131s\u0131. K’nin varsay\u0131lan de\u011feri 2’dir, dolay\u0131s\u0131yla yaln\u0131zca bir tahmin de\u011fi\u015fkeni olan bir modelin K de\u011feri 2+1 = 3 olacakt\u0131r. ln (L) […]<\/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-4508","post","type-post","status-publish","format-standard","hentry","category-rehber"],"yoast_head":"\n\n
\n
\u00d6rnek: R’de \u00d6zellik Se\u00e7imi i\u00e7in stepAIC() i\u015flevinin kullan\u0131lmas\u0131<\/strong><\/span><\/h2>\n
#view first six rows of mtcars dataset<\/strong><\/span>\nhead(mtcars)\n\n mpg cyl disp hp drat wt qsec vs am gear carb\nMazda RX4 21.0 6 160 110 3.90 2.620 16.46 0 1 4 4\nMazda RX4 Wag 21.0 6 160 110 3.90 2.875 17.02 0 1 4 4\nDatsun 710 22.8 4 108 93 3.85 2.320 18.61 1 1 4 1\nHornet 4 Drive 21.4 6 258 110 3.08 3.215 19.44 1 0 3 1\nHornet Sportabout 18.7 8 360 175 3.15 3.440 17.02 0 0 3 2\nValiant 18.1 6 225 105 2.76 3,460 20.22 1 0 3 1\n<\/strong><\/pre>\n\n
library<\/span> (MASS)\n\n#fit initial multiple linear regression model\n<\/span>model <- lm(hp ~ mpg + wt + drat + qsec, data=mtcars)\n\n#use both forward and backward selection to find model with lowest AIC\n<\/span>stepAIC(model, direction=\" both<\/span> \")\n\nStart: AIC=226.88\nhp ~ mpg + wt + drat + qsec\n\n Df Sum of Sq RSS AIC\n- drat 1 94.9 28183 224.98\n- mpg 1 1519.4 29608 226.56\n none 28088 226.88\n- wt 1 3861.9 31950 229.00\n-qsec 1 28102.2 56190 247.06\n\nStep: AIC=224.98\nhp ~ mpg + wt + qsec\n\n Df Sum of Sq RSS AIC\n- mpg 1 1424.5 29608 224.56\n none 28183 224.98\n+ drat 1 94.9 28088 226.88\n- wt 1 3797.9 31981 227.03\n-qsec 1 29625.1 57808 245.97\n\nStep: AIC=224.56\nhp ~ wt + qsec\n\n Df Sum of Sq RSS AIC\n none 29608 224.56\n+ mpg 1 1425 28183 224.98\n+ drat 1 0 29608 226.56\n- wt 1 43026 72633 251.28\n-qsec 1 52881 82489 255.35\n\nCall:\nlm(formula = hp ~ wt + qsec, data = mtcars)\n\nCoefficients:\n(Intercept) wt qsec \n 441.26 38.67 -23.47 \n<\/strong><\/span><\/pre>\n Ek kaynaklar<\/strong><\/span><\/h2>\n
R’de par\u00e7al\u0131 regresyon nas\u0131l ger\u00e7ekle\u015ftirilir?<\/a>
R’de spline regresyonu nas\u0131l ger\u00e7ekle\u015ftirilir<\/a><\/span><\/p>\n","protected":false},"excerpt":{"rendered":"