{"id":1228,"date":"2023-07-27T05:25:49","date_gmt":"2023-07-27T05:25:49","guid":{"rendered":"https:\/\/statorials.org\/tr\/olceklenecek-konum\/"},"modified":"2023-07-27T05:25:49","modified_gmt":"2023-07-27T05:25:49","slug":"olceklenecek-konum","status":"publish","type":"post","link":"https:\/\/statorials.org\/tr\/olceklenecek-konum\/","title":{"rendered":"\u00d6l\u00e7ek ve konum grafi\u011fi nas\u0131l yorumlan\u0131r: \u00f6rneklerle"},"content":{"rendered":"

<\/p>\n

\n

\u00d6l\u00e7eklendirilmi\u015f konum grafi\u011fi,<\/strong> x ekseni boyunca bir regresyon modelinin<\/a> uygun de\u011ferlerini ve y ekseni boyunca standartla\u015ft\u0131r\u0131lm\u0131\u015f art\u0131klar\u0131n karek\u00f6k\u00fcn\u00fc g\u00f6r\u00fcnt\u00fcleyen bir grafik t\u00fcr\u00fcd\u00fcr.<\/span> <\/p>\n

\"\u00d6l\u00e7eklendirilmi\u015f<\/p>\n

Bu grafi\u011fe bakarak iki \u015feyi kontrol ediyoruz:<\/span><\/p>\n

1.<\/strong> K\u0131rm\u0131z\u0131 \u00e7izginin \u00e7izim \u00fczerinde yakla\u015f\u0131k olarak yatay oldu\u011funu do\u011frulay\u0131n. E\u011fer durum buysa, o zaman e\u015f varyans<\/a> varsay\u0131m\u0131 belirli bir regresyon modeli i\u00e7in muhtemelen kar\u015f\u0131lan\u0131r. Yani art\u0131klar\u0131n da\u011f\u0131l\u0131m\u0131 t\u00fcm uygun de\u011ferler i\u00e7in yakla\u015f\u0131k olarak e\u015fittir.<\/span><\/p>\n

2.<\/strong> Art\u0131klar aras\u0131nda net bir trend olup olmad\u0131\u011f\u0131n\u0131 kontrol edin. Ba\u015fka bir deyi\u015fle, art\u0131klar, t\u00fcm uyum de\u011ferleri i\u00e7in yakla\u015f\u0131k olarak e\u015fit de\u011fi\u015fkenlik ile k\u0131rm\u0131z\u0131 \u00e7izgi etraf\u0131nda rastgele da\u011f\u0131lmal\u0131d\u0131r.<\/span><\/p>\n

R’de \u00f6l\u00e7ek ve konum \u00e7izimi<\/strong><\/span><\/h3>\n

R’de basit bir do\u011frusal regresyon modelini s\u0131\u011fd\u0131rmak ve ortaya \u00e7\u0131kan model i\u00e7in bir \u00f6l\u00e7ek ve konum grafi\u011fi olu\u015fturmak i\u00e7in a\u015fa\u011f\u0131daki kodu kullanabiliriz:<\/span> <\/p>\n

 #fit simple linear regression model\n<\/span>model <- lm(Ozone ~ Temp, data = airquality)\n\n#produce scale-location plot\n<\/span>plot(model)\n<\/strong><\/pre>\n
\"R'de<\/p>\n
 Bu regresyon modelinin \u00f6l\u00e7ek-konum grafi\u011finden a\u015fa\u011f\u0131daki iki \u015feyi g\u00f6zlemleyebiliriz.<\/span><\/p>\n
 1.<\/strong> K\u0131rm\u0131z\u0131 \u00e7izgi arsa \u00fczerinde yakla\u015f\u0131k olarak yatayd\u0131r. E\u011fer durum b\u00f6yleyse, o zaman belirli bir regresyon modeli i\u00e7in e\u015f varyans<\/a> varsay\u0131m\u0131 kar\u015f\u0131lan\u0131r. Yani art\u0131klar\u0131n da\u011f\u0131l\u0131m\u0131 t\u00fcm uygun de\u011ferler i\u00e7in yakla\u015f\u0131k olarak e\u015fittir.<\/span><\/p>\n
 2.<\/strong> Art\u0131klar aras\u0131nda net bir trend olup olmad\u0131\u011f\u0131n\u0131 kontrol edin. Ba\u015fka bir deyi\u015fle, art\u0131klar, t\u00fcm uyum de\u011ferleri i\u00e7in yakla\u015f\u0131k olarak e\u015fit de\u011fi\u015fkenlik ile k\u0131rm\u0131z\u0131 \u00e7izgi etraf\u0131nda rastgele da\u011f\u0131lmal\u0131d\u0131r.<\/span><\/p>\n
\n
 Teknik notlar<\/strong><\/span><\/p>\n
 Veri k\u00fcmesinden en y\u00fcksek standartla\u015ft\u0131r\u0131lm\u0131\u015f art\u0131klara sahip \u00fc\u00e7 g\u00f6zlem grafikte etiketlenmi\u015ftir.<\/span><\/p>\n
 30, 62 ve 117. sat\u0131rlardaki g\u00f6zlemlerin en y\u00fcksek standartla\u015ft\u0131r\u0131lm\u0131\u015f art\u0131klara sahip oldu\u011funu g\u00f6rebiliriz.<\/span><\/p>\n
 Bu, bu g\u00f6zlemlerin mutlaka ayk\u0131r\u0131 de\u011ferler oldu\u011fu anlam\u0131na gelmez, ancak bu g\u00f6zlemleri daha yak\u0131ndan incelemek i\u00e7in orijinal verilere bakmak isteyebilirsiniz.<\/span><\/p>\n<\/blockquote>\n
 \u00d6l\u00e7ek konum grafi\u011finde k\u0131rm\u0131z\u0131 \u00e7izginin kabaca yatay oldu\u011funu g\u00f6rebilsek de, bu yaln\u0131zca e\u015f varyans varsay\u0131m\u0131n\u0131n kar\u015f\u0131lan\u0131p kar\u015f\u0131lanmad\u0131\u011f\u0131n\u0131 g\u00f6rmenin g\u00f6rsel bir yolu olarak hizmet eder.<\/span><\/p>\n
 E\u015f varyans varsay\u0131m\u0131n\u0131n kar\u015f\u0131lan\u0131p kar\u015f\u0131lanmad\u0131\u011f\u0131n\u0131 g\u00f6rmek i\u00e7in kullanabilece\u011fimiz resmi bir istatistiksel test Breusch-Pagan testidir<\/a> .<\/span><\/p>\n
 R’de Breusch-Pagan testi<\/strong><\/span><\/h3>\n
 A\u015fa\u011f\u0131daki kod, R’de Breusch-Pagan testi ger\u00e7ekle\u015ftirmek i\u00e7in lmtest<\/strong> paketinin bptest()<\/strong> i\u015flevinin nas\u0131l kullan\u0131laca\u011f\u0131n\u0131 g\u00f6sterir:<\/span><\/p>\n
 #load lmtest package\n<\/span>library(lmtest)\n\n#perform Breusch-Pagan Test\n<\/span>bptest(model)\n\n\tstudentized Breusch-Pagan test\n\ndata: model\nBP = 1.4798, df = 1, p-value = 0.2238\n<\/strong><\/pre>\n
 Breusch-Pagan testi a\u015fa\u011f\u0131daki bo\u015f ve alternatif hipotezleri kullan\u0131r:<\/span><\/p>\n