Yazar "Sanisoglu S.Y." seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • [ X ]
    Öğe
    The criteria for classification tree methods in clinical researches
    (2010) Akkus Z.; Sanisoglu S.Y.; Ugurlu M.; Celik M.Y.
    This study aimed at evaluating a statistical method, classification tree, which are recently developed parallel to the improvements in computer technology. The advantages over other methods and the criterions developed for classification tree are reported in this study. Classification tree (CT) is a non-parametric statistical method using a tree algorithm for reaching diagnosis by utilizing one or more risk factors. Classifications (discriminative, logistic regression and cluster analysis etc) and regression methods are frequently employed in analysing data acquired from scientific studies. However, hypothesis in these models makes the statistical analysis limited to be performed in wide range of disciplines. As there is no need for hypothesis in analysing these data sets, classification trees are serious alternative for other statistical classification and regression techniques. Classification tree, also known as Decision tree, is a good choice for data mining classifications in respect to both understanding and explaning the some particular rules about estimating the results. These methods are evolved following the improvements in computer technology. Classification tree is becoming more important in practice as it provides reliable measures in building accurate classifications. The advantages of the method over others are the following: simplification of the results, provision of non-parametric and lineer solutions, generalization of the conclusions optained by inductive reasoning. More over the technique can utilize mixed data types and the same variable can be employed in different parts of the tree. The determination of choices, which is crucially important in accurate interpretation of the results, needs time and effort in practicing the method. In field of medicine, classification tree is one of the favorable methods particulary utilized in clinical studies.
  • [ X ]
    Öğe
    Determination of medicine and dentistry students? Market choice and their attitudes in preferring brand name products using binary logistics regression method
    (2012) Akkus Z.; Satici O.; Sanisoglu S.Y.; Akyol M.; Keskin S.
    The aim of this study is to determine Medicine and Dentistry students' market choices and their attitude in buying brand name products. Currently, price and quality of products in hypermarkets are the common features consumers care when shopping. It is well known that many people prefer cheap and quality products and maintain a relative attitude due to increasing economic difficulties. 229 medicine and 222 dentistry (totally 451) students comprise the samples of this study. A poll of 28 questions was conducted to determine the preferences of the students in shopping. Taking market preferences as dependent variable, and other related variables as independent, the parameters affecting market preferences are analyzed with Binary Logistic Regression Method. To the question "Which market do you prefer in shopping?", 155 students (%34,4) replied BIM, 138 of them (%30.8) replied Grocer's, 88 of them (%19.5) replied Migros, 70 of them (%15.5) said Carrefoursa is the market they prefer in shopping. As result of the survey, in the first and second analysis, correct classification rates have been found to be 81.6% and 76.3% respectively. It has been found out that the most effective parameters in preferring markets for shopping are age, father's education, type of accommodation (house, dorms, with parents, etc.), buying the markets' brand name products, selling quality products, product variety, market's being modern and spacious. In field works, because it produces more reliable results than both univariate logistic analysis and other statistical analysis when the variables representing the data are discrete variables, multivariate logistic regression analysis is considered to be the best method to be used.