<\/p>\n
Bir modelin tahmin do\u011frulu\u011funu \u00f6l\u00e7mek i\u00e7in en yayg\u0131n kullan\u0131lan \u00f6l\u00e7\u00fcmlerden biri, a\u011f\u0131rl\u0131kl\u0131 ortalama mutlak y\u00fczde hatas\u0131<\/strong> anlam\u0131na gelen WMAPE’dir<\/strong> .<\/span><\/p>\n WMAPE’yi hesaplamak i\u00e7in form\u00fcl:<\/span><\/p>\n WMAPE<\/strong> = (\u03a3|y i<\/sub> \u2013 \u0177 i<\/sub> |*w ben<\/sub> ) \/ (\u03a3y i<\/sub> *w ben<\/sub> ) * 100<\/span><\/p>\n Alt\u0131n:<\/span><\/p>\n R’de WMAPE’yi hesaplamak i\u00e7in a\u015fa\u011f\u0131daki i\u015flevi tan\u0131mlayabiliriz:<\/span><\/p>\n A\u015fa\u011f\u0131daki \u00f6rnekte bu fonksiyonun pratikte nas\u0131l kullan\u0131laca\u011f\u0131 g\u00f6sterilmektedir.<\/span><\/p>\n Bir perakende ma\u011fazas\u0131n\u0131n fiili ve tahmini sat\u0131\u015flar\u0131 hakk\u0131nda bilgi i\u00e7eren R’de a\u015fa\u011f\u0131daki veri \u00e7er\u00e7evesine sahip oldu\u011fumuzu varsayal\u0131m:<\/span><\/p>\n Ger\u00e7ek ve tahmini sat\u0131\u015flar aras\u0131ndaki farka ili\u015fkin WMAPE’yi hesaplamak i\u00e7in kullan\u0131lacak bir a\u011f\u0131rl\u0131k vekt\u00f6r\u00fc tan\u0131mlayabilir, ard\u0131ndan daha \u00f6nce tan\u0131mlad\u0131\u011f\u0131m\u0131z WMAPE i\u015flevini kullanabiliriz:<\/span><\/p>\n Bu modelin WMAPE’si %13,27635<\/strong> olarak \u00e7\u0131k\u0131yor.<\/span><\/p>\n Yani tahmin edilen sat\u0131\u015f de\u011ferleri ile ger\u00e7ekle\u015fen sat\u0131\u015f de\u011ferleri aras\u0131ndaki a\u011f\u0131rl\u0131kl\u0131 ortalama mutlak y\u00fczdelik hata %13,27635’tir.<\/span><\/p>\n Bu \u00f6rnekte Ocak ve \u015eubat de\u011ferlerine \u00e7ok daha b\u00fcy\u00fck a\u011f\u0131rl\u0131klar atad\u0131\u011f\u0131m\u0131z\u0131 unutmay\u0131n.<\/span><\/p>\n \u00d6zel probleminize ba\u011fl\u0131 olarak, modelinizdeki her hatan\u0131n \u00f6nemine g\u00f6re farkl\u0131 g\u00f6zlemlere daha fazla veya daha az a\u011f\u0131rl\u0131k atayabilirsiniz.<\/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 MAPE nas\u0131l hesaplan\u0131r<\/a> Bir modelin tahmin do\u011frulu\u011funu \u00f6l\u00e7mek i\u00e7in en yayg\u0131n kullan\u0131lan \u00f6l\u00e7\u00fcmlerden biri, a\u011f\u0131rl\u0131kl\u0131 ortalama mutlak y\u00fczde hatas\u0131 anlam\u0131na gelen WMAPE’dir . WMAPE’yi hesaplamak i\u00e7in form\u00fcl: WMAPE = (\u03a3|y i \u2013 \u0177 i |*w ben ) \/ (\u03a3y i *w ben ) * 100 Alt\u0131n: \u03a3 \u2013 \u201ctoplam\u201d anlam\u0131na gelen bir sembol y i \u2013 i’inci g\u00f6zlemin […]<\/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-3967","post","type-post","status-publish","format-standard","hentry","category-rehber"],"yoast_head":"\n\n
find_WMAPE <- function<\/span> (y, yhat, w){\n return<\/span> (sum(abs(y-yhat)*w)\/sum(y*w)*100)\n}\n<\/strong><\/pre>\n \u00d6rnek: R’de WMAPE’nin hesaplanmas\u0131<\/strong><\/span><\/h2>\n
#create dataset<\/span>\ndata <- data. frame<\/span> (actual=c(23, 37, 44, 47, 48, 48, 46, 43, 32, 27, 26, 24),\n forecast=c(37, 40, 46, 44, 46, 50, 45, 44, 34, 30, 22, 23))\n\n#view dataset<\/span>\ndata\n\n current forecast\n1 23 37\n2 37 40\n3 44 46\n4 47 44\n5 48 46\n6 48 50\n7 46 45\n8 43 44\n9 32 34\n10 27 30\n11 26 22\n12 24 23\n<\/strong><\/pre>\n #define function to calculate WMAPE\n<\/span>find_WMAPE <- function<\/span> (y, yhat, w){\n return<\/span> (sum(abs(y-yhat)*w)\/sum(y*w)*100)\n}\n\n#define weights for each month\n<\/span>weights <- c(20, 20, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6)\n\n#calculate WMAPE\n<\/span>find_WMAPE(df$actual, df$predicted, weights)\n\n[1] 13.27635\n<\/strong><\/pre>\n Ek kaynaklar<\/strong><\/span><\/h2>\n
R’de SMAPE nas\u0131l hesaplan\u0131r<\/a>
R’de RMSE nas\u0131l hesaplan\u0131r<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"