LEE- Endüstri Mühendisliği-Yüksek Lisans
Bu koleksiyon için kalıcı URI
Gözat
Son Başvurular
1 - 3 / 3
-
ÖgeMakina öğrenmesi teknikleri ile hukuki alacak tahsilat kuruluşu dosya kapatılabilirlik tahmini ve atama modeli ile dosya ataması: Telekomünikasyon sektörü örneği(Lisansüstü Eğitim Enstitüsü, 2022-06-28)Günümüzde, bütün sektörlerde olduğu gibi telekomünikasyon sektöründe de görülen rekabetçi yaklaşım, ilgili sektörde faaliyet gösteren şirketlerin yeni müşteriler kazanmasını oldukça zor bir hale getirmektedir. Durum böyle iken bu şirketler makine öğrenmesi yöntemleri ile müşteriye özel kampanyalar sunarak, toplam müşteri sayılarını artırmaya veya halihazırdaki müşterileri kendi markalarında tutmaya çalışmaktadırlar. Ancak makine öğrenmesi yöntemlerinin uygulama alanı sadece müşteri kazanma veya var olan müşterinin sadakatini sağlamaya yönelik çalışmalardan ibaret değildir. Şirketlerin büyümeye devam edebilmesi için kazanılan müşterilerden kazanç elde etmeleri gerekmektedir. Eğer şirketin halihazırda sahip olduğu müşteriler, aldığı hizmet karşılığında şirkete olan yükümlülüklerini yerine getiremiyor ise, şirketin bu gibi müşterileri kazanmak için sarf ettiği çaba, şirkete karlılık değil aksine yük getirmektedir. Bu yük sadece müşterilerden elde edilecek gelirin karşılanamaması olarak düşünülmemelidir. Müşteriden gelir elde edememenin yanında, müşteriye ait borcun tahsilatı için şirketin ayıracağı kaynaklar da göz önünde bulundurulmalıdır. Bu gibi durumlarda şirketler hem çalışanlarını hem de diğer şirket kaynaklarını (gerekli teknik ekipmanlar, elektrik vb.) bu kanala bağlayarak zarara uğrayacaktır. Şirketler, müşteriden borç tahsilatı yapabilmek için ya kurum içi kaynaklarını kullanmakta ya da kurum dışı kaynaklar aracılığıyla bu ödenmemiş borcu tahsil etmektedir. Burada bahsi geçen dış kaynaklar, alacak tahsilat kuruluşlarıdır ve bu kuruluşlar müşterinin ilgili şirkete olan borcuna ilave olarak, kendi verdiği hizmetin karşılığı olan tutarı da ekleyerek şirket müşterisinden ödenmemiş borcu tahsil etmeye çalışmaktadır. Şirketler tarafından ödenmemiş borçları tahsil etmeye yönelik yapılan çalışmalar büyük ölçüde müşterilerin ödeme istekliliğine ve alacak tahsilat kuruluşlarının müşteri ile müzakerelerine bağlı olan alacak tahsilat süreçleridir. Bu süreçte, tahsildarlar önce borçlu müşterileri borçlarını ödemeye ikna etmeye çalışmakta eğer borçlu müşteriler, borçlarını bu müzakereler sonucunda belirtilen süre içerisinde ödeyemezler ise yasal alacak tahsilatı süreci başlamaktadır. Yasal alacak tahsili aşamasında, alacak davalarını yargı yoluyla çözmeye çalışan, alanında tecrübeli avukatların çalıştığı hukuk büroları mevcuttur. Büyük şirketler, borçlarını ödeyemeyen müşterilerinden borç tahsilatı yapmak amacıyla iç kaynaklarını genişletmek yerine, yasal alacak tahsilat kuruluşları olarak sözleşmeli hukuk büroları ile çalışmaktadırlar. Bu tez kapsamında, çalışmanın yapıldığı ilgili telekomünikasyon şirketi de müşterileri tarafından ödenmemiş faturaların tahsilatını, dış kaynak olarak, anlaşmalı bulunduğu hukuk büroları aracılığıyla sağlamaktadır. Şirket, borcunu ödeyemeyen ve yasal alacak takibine girmiş müşterilerinden, anlaşmalı bulunduğu hukuk büroları aracılığıyla alacak tahsilatı yapmaya çalışmaktadır ve bu rekabetçi ortamda şirketin ayakta kalabilmesi için alacak tahsil etme sürecini en iyi şekilde yönetmesi gerekmektedir. Bu aşamada anlaşmalı hukuk büroları, büronun performans ve tecrübesinin alacak davalarını kapatıp kapatamaması üzerindeki etkisine göre değerlendirilmektedir. Hukuk bürolarının performansının belirlenmesindeki amaç, büronun amaçlarına ne ölçüde ulaştığını objektif bir şekilde ve periyodik olarak ölçümlemektir. Performans ölçümü; ilgili hukuk bürosunun girdilerini, iç süreçlerini, prosedürlerini, çıktılarını ve sonuçlarını değerlendirebilmeyi sağlamaktadır. Böylece bu performans ölçümü sonucu göstergeler analiz edilerek doğru karar verilebilmekte, sürekli iyileştirme yaratabilecek sınırlı kaynaklar etkin bir şekilde geliştirilebilmekte ve bireylere hedefler verilebilmektedir. Bu çalışmanın yapıldığı şirkete ait geçmiş veri setlerinden hareketle, şirketin anlaşmalı olduğu hukuk bürolarına devredilmiş ve başarılı şekilde tahsilat sağlanmış alacak davalarının olmasının yanı sıra dava dosyasının devredildiği büro tarafından tahsilatın sağlanamadığı alacak davaları da olabilmektedir. Hangi dava dosyasının hangi hukuk bürosuna devredildiği, şirketin alacaklarını hukuki aşamada tahsil edebilmesi açısından kritik bir öneme sahiptir. Bu doğrultuda, bu çalışma, ilk aşamada geçmiş tarihsel verilere dayalı makina öğrenmesi yöntemlerini kullanarak ilgili şirket tarafından anlaşmalı olunan hukuk bürolarının alacak dava dosyası kapatma olasılıklarını tahminleyerek, devredilecek dosyayı daha çok kapatma kabiliyetine sahip hukuk bürosunu bulmayı amaçlamaktadır. Dava dosyası kapatma olasılığını tahmin etmek için, işlenen veri kümesine 9 makina öğrenme algoritması; Catboost Sınıflandırıcı, Extreme Gradient Boost Sınıflandırıcı, Gradient Boosting Sınıflandırıcı vb. uygulanmıştır. Sonuçlar, Catboost Sınıflandırıcı'nın 0,917 doğruluk oranı ile en iyi doğruluk performansına sahip olduğunu göstermiştir. Ayrıca, boosting tipi topluluk öğrenme algoritmalarının diğer algoritmalardan daha iyi performans gösterdiği sonucuna varılmıştır. Son olarak, modellemede daha iyi doğruluk elde etmek için Catboost Sınıflandırıcısının hiper parametreleri ayarlanmış ve hiper parametre optimizasyonu, tahmin üzerinde %0,01 oranda olumlu etki sağlamıştır. Son olarak modelin test kararlılığını test etmek için k-katlı çapraz doğrulama yapılmıştır. Makine öğrenimi modelinde seçilen CatBoost algoritması üzerinden 2021 Haziran ayına ait takip dosyalarının, anlaşmalı hukuk büroları tarafından kapatılabilme olasılıkları tahminlenmiştir. Bu kapatılabilirlik puanları, çalışmanın ikinci aşamasında, kurulan kapatılabilirlik temelli adil dosya atama modeline girdi olarak sağlanmıştır. Bu bir optimizasyon modeli olup, bu modelde amaç, takip dosyalarının tahsilatla kapatılma durumunu en büyüklerken, bütçe kısıtı altında anlaşmalı hukuk bürolarına eşit bir şekilde takip dosyalarının atamalarını gerçekleştirmektir. Kapatılabilirlik temelli adil dosya atama modeli ile 2021 Haziran ayına ait 33,365 adet takip dosyasının, anlaşmalı bulunulan 102 hukuk bürosuna optimum bir şekilde atamaları gerçekleştirilmiş olup sonuçlar üzerinden kurulan modeller değerlendirilmiştir. Atama çıktılarının büyük bir oranda modelin amaçlarına ve kısıtlarına uygun şekilde yapıldığı sonucuna varılmıştır. Bunun üzerine önce makine öğrenimi modelinin çalıştığı ardından makine öğrenimi modeli çıktısı olan dosya kapatılabilirlik değerlerinin atama modeline beslendiği ve atama modelinin çözdürülüp gerekli atama sonuçlarının alındığı tümüyle otomatik bir şekilde çalışacak olan yasal alacak takibi süreci karar destek sistemi kurulmuştur
-
ÖgeRanking of families applying for social aids to municipalities according to their degree of neediness(Graduate School, 2022-06-20)Although there is no general definition of poverty, it varies according to time and place. The concept of poverty, which is intertwined with many social variables, can be stated as being deprived of minimum living standards in general. Although there are many sub-titles of poverty, the concepts of absolute deprivation and relative deprivation are the most mentioned concepts in the literature. Absolute deprivation refers to the inability of households or individuals to meet the basic needs for their survival, while relative deprivation refers to being below the average welfare level created in the society. Various policies have been developed globally and nationally in the fight against poverty. One of the most important international common approaches is the United Nations Sustainable Development Goals (SDGs). This call to action, also known as the Global Goals; aims to protect our planet, eliminate poverty, and ensure that all people live in peace and prosperity. The legal dimension of the social security and social assistance system in Turkey has been drawn by laws and the constitution. The Social Services Law of 1983 regulates the principles and procedures regarding the social services taken to individuals in need of protection, assistance and/or care, and the authorities, duties and responsibilities of the institutions established to carry out these services. The law also states that if the service demand is more than the service supply, the priority will be determined by the degree of neediness and the order of application/detection. The General Directorate of Social Assistance and Solidarity was established in 2004, and in 2005 it was given the task of providing social assistance to special provincial administrations and municipalities. This decision has a critical importance because municipalities are important administrative units that provide the closest service to the citizens and have a direct impact on their lives. Every year, municipalities allocate a part of their annual budget to social aids. These social aids can be form of food, clothing, fuel, etc. or it can also be done directly in cash. The COVID-19 pandemic, which has started in 2019, affected both local and global economies, in terms of production, employment, social services, health, etc. There have been declines in many sectors. Turkey has also had its share from these economic fluctuations, and according to the statement of the TURKSTAT (Turkish Statistical Institute), annual inflation in 2021 was 36.08%. According to the research carried out by Türk-İş Union, in December 2021, the hunger treshold for a family of four in Turkey was determined as 4,013.26 TL (Turkish Lira) and the poverty line as 13,073 TL while in the meantime the minimum wage was 2,825 TL. Due to the above-mentioned conditions, applications for aid to municipalities have increased in the recent period. For instance, more than 1 million families applied for aid to the Istanbul Metropolitan Municipality during the pandemic period. The necessity of making a ranking for such a large number of applications is quite obvious. The aim of the study is to rank the families applying for social assistance according to their neediness with an easily applicable and consistent formula. In this context, to contribute to the effective use of the budget of the municipalities and to contribute to the literature. Fuzzy Analytic Hierarchy Process and Simple Additive Weighting methods were used for the formula. Analytic Hierarchy Process is a multi-criteria decision-making method introduced by Thomas L. Saaty in 1980. This method, which is based on pairwise comparisons, considers the problem (goal) as a hierarchical structure. In addition to its simplicity and usefulness, one of the most important reasons underlying the popularity of the method is that it can use qualitative and quantitative factors together. Although AHP's purpose is to capture expert knowledge, it is currently unable to reflect human thinking styles. As a result, F-AHP was developed as a fuzzy extension of AHP to overcome these issues. F-AHP uses the fuzzy set theory that was introduced by Lotfi A. Zadeh in 1965. According to classical set theory, the degree of belonging of an element to a set is 0 or 1. That is, this element either belongs to the set or it does not. In a fuzzy set, the degree of belonging of the elements to the set is defined by the "membership value" and varies between 0 and 1. One of the most important advantages of the F-AHP is that it can benefit from linguistic terms in pairwise comparisons. When comparing two criteria, the decision maker can make healthier comparisons with linguistic terms such as "more important", "much more important", "equally important" rather than giving crisp values. Simple Additive Weighting Method (SAW) is one of the Multi Attribute Decision Making (MADM) method introduced by Ackoff and Churchman in 1954. This method is also often known as the weighted summing method. The method is based on the weighted average. For each alternative, an evaluation score is calculated by multiplying the scaled value given to that attribute's alternative with the weights of relative importance directly assigned by the decision maker, then summing the products for all criteria. The criteria required to create the ranking formula that we have suggested were obtained by examining the social assistance regulations of various municipalities in Turkey, since there is no former study on the subject in the literature. Social assistance regulations of 27 municipalities, 14 of which are metropolitan municipalities and 13 of which are district municipalities, were examined. As a result of these examinations, the criteria were divided into two headings, economic and demographic.The sub- criteria of the economic criteria were determined as the monthly net income of the household, the amount of monthly aid received from different institutions, movable property insurance value, immovable property fair value, and the monthly rent paid by the family. While the sub-criteria of demographic criteria were determined as the number of people in the family, the number of people working in the family, the number of students in the family, the number of people with disabilities in the family, and the number of people over 70 years old in the family. The problem has been transformed into a hierarchical structure in accordance with the F-AHP. After this process, a questionnaire form was created for pairwise comparisons. The decision makers to fill out the questionnaires were determined as social service experts, sociologists, psychologists in municipalities and people in the municipality administration. In the study, which was carried out as an online questionnaire, the form was sent in Turkish, and a version translated into English was added to the study. In the study, 26 questionnaires were analyzed. The results obtained were analyzed in overall, only for metropolitan municipalities and district municipalities separately, and criteria weights were calculated. In all three scenarios, economic criteria were determined to be more important than demographic criteria. The most important among the 10 sub-criteria was the monthly net income of the household. Since families with higher scores are described as more needy in the formula created, amongst the sub-criteria, the sufficiency and deficiency criteria are determined. The normalization of these two different types of criteria will be performed differently from each other. The determination of the criteria, the survey process, the derivation of the criteria weights, and the decomposition of the criteria for normalization can be found in section 4. In section 5, the created formula is shown through a numerical example. In the example, a 100x10 value matrix was created with the answers of 100 different families assumed to have applied to 10 criteria. After eliminating those who were not in need out of 100 families in the first step, the values were normalized for the remaining 85 families and the score of each family was calculated separately. Families in the top 10 were determined. Operations were performed for the three different weightings obtained. Finally, in section 6, a sensitivity analysis was performed for reliability and consistency.
-
ÖgeConceptualisation and instrument development for mobile application usability based on android operating system(Lisansüstü Eğitim Enstitüsü, 2021)Notions such as mobile internet and mobile applications ceased to be a nascent concept when basic cell phones, which became prevalent in the society in the late 1990s, left their dominations to smartphones in the early 2010s. The technological infrastructure, which has made substantial progress in the framework of mobile concept, the organisational models of the companies and the penetration rate of the mobile internet among users exceeding the usage rate of desktop internet, has significantly speeded up the use of smartphones and mobile applications that may be used anywhere and anytime. Large-scale usage of mobile applications is based on companies developing mobile applications and directing their services to the context of mobile applications to remain firm in the competitive market. In this perspective, the Google Play Store and Apple App Store platforms, where mobile applications are delivered to users by Android and iOS operating systems respectively, stand out as an important figure with millions of different applications. The large number of mobile applications has given users the opportunity to turn to lots of different options. Hence, it has become requisite for companies to gain advantage in a strong competitive environment by better marketing their products and services in the digital environment in order to continue their viability. One of the essential points of ensuring adoption, preferability, and continuity in such a rapidly growing mobile application universe is that mobile applications can be used in a way that makes user experience better. Yet, when the literature was scrutinised well, it was noticed that mostly desktop computer or website usability models were adapted to the mobile device and application context or mobile application usability was tried to be evaluated with very usual criteria such as satisfaction, learnability, memorability, effectiveness, and effectiveness. It is crystal clear that the usability concept, which plays a more vital role in mobile technology, should not be addressed with the evaluation methods of desktop computers and these methods should not be directly adapted to the mobile device or mobile application concept. Therefore, mobile devices and particularly mobile applications with which users interact directly should have their own usability evaluation criteria. However, the lack of such a usability model for the Android operating system, which is one of the two largest operating systems and is currently preferred by roughly 4 out of every 5 smartphone user, is a major gap realised in the literature. Hereby, a comprehensive and conceptualised usability model for mobile applications running on the Android operating system should be established. In the literature part of this research, mobile technologies, mobile shopping applications that will be the subject of the analysis phase of the study, and the concept of usability are explained meticulously. A conceptual model for mobile application usability running on the Android operating system was established and verified with a survey instrument. Specifically, the guide for interface design of Android-based mobile applications, published by Google, was thoroughly analysed and adapted to the conceptual model. Then, in the light of this reviewed guide, 13 constructs illustrating the usability of mobile applications running on the Android operating system were originated. Open and axial coding was done following the methodology. The matrix was designed by listing these codes, and used in the next stage of this study. In the second phase of the methodology of this thesis, firstly items representing each construct within the context of the matrix were created and collected in the item pool. These items were first evaluated by experts and then by the end user, and items that were confusing, ambiguous and meaningless were removed from the pool or modified. In order to develop the scale, face validity check (pre-test), pilot test (n = 30) and content validity check (n = 41) were carried out, respectively. By virtue of these checks, a qualitative and quantitative (using PSA and CSV indices) evaluation of the scale was made. The study was strengthened by gathering feedback about the scale from those who voluntarily participated in these survey studies, and the corrections were made in the item pool thanks to the feedback received. The survey method was used in the analysis stage of this study. People using mobile shopping applications were selected as a target sample. In the analysis part, items in the item pool were adapted to the mobile shopping application concept. Several grounds for the analysis based on mobile shopping applications: increasing mobile shopping volume and popularity because of the Covid-19 pandemic, the reason some of the most downloaded applications in Google Play Store and Apple App Store in Turkey are shopping applications, large investments in the retail sector, policies of companies that direct their customers to mobile shopping and convenience and comfort factors with respect to customers. For exploratory and confirmatory factor analyses, two large samples (n1 = 293 and n2 = 340) were picked and factor, convergent and discriminant validation was fulfilled using IBM SPSS Statistics 26 and IBM SPSS AMOS 26 software. In addition, Cronbach's alpha values of the factors were calculated in both analysis stages and reliability was also measured. During the confirmatory factor analysis phase, nomological validation was made. For nomological validation, 13 constructs were taken into account as independent variables and examined how they affect the variables of brand loyalty, continued intention to use and satisfaction, which are frequently associated with usability in the literature. According to the analysis results, it was proved that the usability model created for Android-based applications has a significant effect on satisfaction, continued intention to use and brand loyalty and predicts these variables decently. Moreover, in the evaluation phase of both factor analysis, the fact that the loadings of the items included in the factors were above the required threshold approved this thesis. As a result, this study is very significant for practitioners working in the field of mobile applications. It will be a guide for mobile application developers and companies working in this sector. The created mobile application usability conceptualisation will allow the use of the practitioners to develop new mobile applications or to enhance an existing application. Therefore, in the last part of the study, the implications of the study and its theoretical contributions for practitioners are mentioned, besides.