Hayat, Elvan AktürkAlpay, Olcay2025-03-232025-03-2320131303-63192791-7614https://hdl.handle.net/11486/2469Data Envelopment Analysis (DEA), a method commonly used to measure the efficiency is becoming an increasingly popular management tool. On the contrary to classical efficiency approaches, the most important advantage of DEA is that researchers can determine the weight restrictions of input and output variables. Variable selection and determination of weight restrictions are important issues in DEA. This work investigates the use of K-nearest neighbor (KNN) algorithm in the definition of weight restrictions for DEA. With this purpose a new approach based on KNN is proposed. Applications are constructed with empirical and real data sets depending on the specific constraints. Performance scores were calculated for both KNN based restricted and unrestricted DEA models and the results are interpreted.Genellikle etkinlik ölçümünde kullanilan Veri Zarflama Analizi (VZA), popüler bir yönetim araci olmaya baslamistir. Klasik etkinlik yaklasimlarinin tersine, VZA’nin en önemli avantaji, girdi ve çikti degiskenlerinin agirlik kisitlarini arastirmacilarin belirleyebilmesidir. Degisken seçimi ve agirlik kisitlarinin belirlenmesi VZA’da önemli konulardir. Bu çalisma VZA için agirlik kisitlarinin tanimlanmasinda K-en yakin komsuluk algoritmasinin kullanimini arastirmaktadir. Bu amaçla K-en yakin komsuluk temeline dayanan yeni bir yaklasim önerilmistir. Belirlenen kisitlara bagli olarak ampirik ve gerçek veri setleri ile uygulamalar yapilmistir. K-en yakin komsu temelinde kisitli model ve agilik kisitlamasiz VZA modeli için performans skorlari hesaplanmistir ve sonuçlar yorumlanmistir.eninfo:eu-repo/semantics/openAccessData envelopment analysisEfficiencyK-nearest neighborWeight restrictionsArticle1036474