İ.T.Ü. Fakültelerinin araştırma etkinlikleri sekreterliğinden yararlanma etkinliklerinin veri zarflama analizi ile belirlenmesi

Abstract

"İstanbul Teknik Üniversitesi Fakültelerinin Araştırma Etkinlikleri Sekreterliğinden Yararlanma Etkinliklerinin Veri Zarflama Analizi İle Belirlenmesi" adlı yüksek lisans tezi giriş bölümü ile birlikte dört ana bölümden oluşmaktadır. Giriş bölümünden sonraki bölümde öncelikle Veri Zarflama Analizi tekniklerinin de içinde bulunduğu Parametrik Olmayan Etkinlik ölçüm yöntemlerinin ortak yönleri anlatılmaktadır. Daha sonra ise bu parametresiz yöntemlerin etkinlik ölçütlerinden söz edilmektedir. Genel olarak parametresiz yöntemler tanımlandıktan sonra özele, Veri Zarflama Analizi yöntemlerine üçüncü bölümde geçilmektedir. Burada BCC, Toplamsal ve CCR Modelleri örnekler ile ayrıntılı bir şekilde anlatıldıktan sonra VZA'nın günümüzdeki açılımları sunulmaktadır. Ardından da VZA'nin uygulama sırasında izlenen temel adımlarından bahsedilmektedir. Daha sonra ise, dördüncü bölümde, uygulama yer almaktadır. Uygulama alam olarak İstanbul Teknik Üniversitesi'nin yeni yönetimince yeniden düzenlenmeye çalışılan Araştırma Ekinlikleri Sekreterliği seçilmiştir. Burada yapılan çalışmalarda, sekreterliğin etkin ve verimli çalışamamasının en büyük nedenlerinden biri olarak üniversite fakültelerinin buradan etkin bir şekilde yararlanamaları olduğu ortaya çıkmıştır. Ayrıca burada çalışanlar ile yapılan görüşmelerden her fakültenin yararlanma durumunun da birbirinden farklı olduğu ortaya çıkmıştır. Bu yüzden uygulama konusu olarak fakültelerin Üniversite Araştırma Etkinlikleri Sekreterliğinden yararlanma etkinlilerinin düzeyinin berlirlenmesi ve aradaki farkların nedenlerinin ortaya çıkarılması seçilmiştir. Gerekli tüm veriler Araştırma Etkinlikleri Sekreterliğinden sonradan girdi ve çıktı ölçütleri olarak ikiye bölünmek üzere, alınmıştır. Daha sonra da geliştirilen model bu verilere uygulanmıştır. Son bölümde ise fakültelerin Araştırma Etkinlikleri Sekreterliğinden yararlanma etkinlikleri arasındaki farklar, farkların görülebilen nedenleri ve öneriler yer almaktadır.

determining the relative efficiencies of the faculties of Istanbul technical university in their relationships with the research activities secretary The thesis contains five main sections. In the first section there is a general introduction to the subject. In the second section, efficiency measurement methods are introduced. They are classified into three groups: 1. Ratio Analysis 2. Parametric Methods 3. Non-Parametric Methods. After a brief introduction about ratio analysis and parametric methods, non- parametric efficiency measurement methods are introduced and Data Envelopment Analysis is put into its place as a non-parametric efficiency measurement method. Initially, the Production Possibility Set, the Observation Set, Observed Input and Output Matrice, the Production Vector, the Efficiency Frontier terms are identified with the help of examples. In the later parts of the section, the efficiency measures are introduced. The Production Possibility Set has 6 kinds according to 6 assumptions. The six categories are explained with a graph. There are four main kinds of efficiency measurents : (i). Technical efficiency (ii). Scale efficiency (iii). Allocative efficiency (iv). Overall efficiency. In the third section the main theme of the thesis comes : Data Envelopment Analysis (DEA). In this part Data Envelopment Analysis is explained especially according to the methods named Additive Method, BCC (by Banker, Charnes and Cooper) and CCR (by Charnes, Cooper and Rhodes). They have some features and the features could be summarized as follows. 1. The Additive model and the extended Additive model (i.) relate DEA to the earlier Chames-Cooper (1959) inefficiency analysis and in the process XIV (ii) relate the efficiency results to the economic concept of Pareto optimality as interpreted in the still earlier work of T. Koopmans (1949). 2. The BCC model (1984) distinguishes between technical and scale inefficiencies by (i) estimating pure technical efficiency at the given scale of operation and (ii) identifying whether increasing, decreasing or constant returns to scale possibilities are present for further exploitation 3. The CCR model (1978) (i) yields an objective evaluation of overall efficiency and (ii) identifies the sources and estimates the amounts of the thus-identified inefficiencies. While each of these models addresses managerial and economic issues and provides useful results, their orientations are different, and more importantly, they generalize and provide contact with these disciplines and concepts. Thus, models may focus on increasing, decreasing, or constant returns to scale as found in economics that are here generalized to the case of multiple outputs. They may determine an efficient frontier that may be piecewise linear, piecewise log-linear, or piecewise Cobb- Douglas with, again, generalization to the output-input situations being achieved in the process. They may utilize non-Archimedian constructs, and they may focus on either input reduction or output augmentation to achieve efficiency. The section seeks to achieve comparisons by focusing on the above basic mathematical models. Primal and dual characterizations for each model are presented, and comparisons between models are developed via geometric portrayals of the corresponding envelopment surfaces, returns to scale properties, projections onto the efficient surface, and invariance of measurement units. In the latter part of the second section additional extentions are discussed. A number of useful enhancements have appeared in the literature, and a comprehensive treatment is beyond the scope of the part. Six extentions analyzed and described are as follows. (i) Nondiscretionary Inputs and Outputs (ii) Categorical Inputs and Outputs (iii) Incorporating Judgement or A Priori Knowledge (iv) Window Analysis (v) Ordinal Inputs and Outputs With Cardinal Inputs and Outputs (vi) Comparative Spatial Disadvantage (CSD) Method of Non-Radial DEA. In the last part of the third section main steps of a DEA application is presented. It has five steps as follows : (i) The determination of observation set (ii) The determination of output-input variable sets (iii) The application of DEA (iv) The Analysis for details (v) Results and comments XV The fourth section has the title "The Application" and it is about the application. Although it is named as application, first of all, the place, The Research Activities Secretary of Istanbul Technical University, that application is performed should be described. As understood from the name, the organization helps the members of Istanbul Technical University in order to make mem research. The fund that they give have seven categories: (i) Scientific Research and Development Supporting Programme (ii) Substructure Supporting Programme For Scientific Research and Development (iii) Young Research People Supprting Programme (iv) Multi-disciplinary Research's supporting Programme (v) Supporting Programme For Researchs About Defense Industry (vi) Supporting Programme For International Cooperation (vii) Graduate and Post Graduate Thesis Supporting Programme After an introduction of the organization, the data that could form production function are tried to collect. However it was so difficult. The organization's stuff could give some data which were yearly. These are; number of projects the faculty members done, the yearly fund the faculty members given, the projects closed, ended, during the year, the number of the projects had final report. But these data are not enough. So the final reports of the projects were read and the reports were tried to convert from words to numbers. Later a survey was done among the stuff of the organization. They were asked about the performances of the faculties. At last the variables listed below were hardly found :. Number of total projects done by the faculty members. Number of projects,that have final reports, done by the faculty members. The yearly funds of the projects. Closed / Total Number of Projects From Refree Reports,. Required literature research is done or not. The Effects of the project to the country. The originality of the project. The consistency between project and its report From Survey,. The knowledge capability of the stuff who concerns about the activities relating with the organization. The knowledge of the project's owner about the registration. The knowledge of the project's owner about the activities done while receiving money. The knowledge of the project's owner about the tracing activities. The speed of the correspondence. The frequency took place during the project. The distance between the organization and the faculty. XVI As it could be seen the variables have different features. Some of them are cardinal, some are categoric, and some are dynamic. Since window analysis will be performed to them, the dynamic variables are separated. Others must be analyzed together, because all are static. In these conditions, a mixed model must be used, thus it is given in the project. When model is chosen, the variables, the data and the decision making units are in thequeu. The variables should be analyzed on behalf of the project. In order to find tiie relationships between them, a kind of hypothesis test, significancy test is applied. Regression analysis is the core of the significancy test. After regression analysis on output and input variables, the following variables of the project are chosen :. Number of total projects done by the faculty members. Number of projects,that have final reports, done by the faculty members. The Effects of the project to the country. The knowledge of the project's owner about the registration. The distance between the organization and the faculty. The first three are outputs and the others are inputs. So the application might begin. The variables belong to the following faculties : Civil, Science & Literature, Chemistry & Metallurgy, Mine, Electric & Electronic, Architecture, Machinary, Aeroplane & Space, Ship Building and Management. After the application two kind of relative efficiencies are found. One of them comes from Window Analysis and it is still dynamic, however the other one is static. As a result of the situation the dynamic one is converted to static by exponential smoothing. Now these efficiencies can be summed up by the proposed weights. A kind of number is found at the end of the analyzes, but it is not known if the weights of the output and input variables chosen are true or not. m order to test it sensitivity analysis should be done. The sensitivity analysis should be done to the following subjects : (i) The model used in the application (ii) The input variables (iii) The output variables (iv) The coefficient that restricts multipliers. XVll In the project it is seen that the sensitivity analysis is very useful. It prevents fee mistakes on the relative efficiencies and the final comments made by the help of these analyses.