Comparing cost sensitive classifiers by the false-positive to false- negative ratio in diagnostic studies

Nowadays researchers want to be cautious about cost of building models which can generate false positives and false negatives in unexpected ways. They keep on searching for various measures for controlling such behavior depending upon the underlying datasets. Cost sensitive classifiers are the models to check the total cost due to misclassifications. In this article, the cost sensitive classifiers are tried for the first time endowed with a new measure 'cost ratio' to monitor such misclassifications in the sensitive diagnostic studies. The scheme for variations of such ratio is introduced and its influence on the loss is investigated. This cost ratio, a rational number, ? is made up of the integers for the cost of false positive by its' frequency of occurrences, in the numerator and the similar cost of false negative in the denominator. We apply this novel cost monitoring measure for learning the sample dataset of sensitive nature in the context of in vitro fertilization (IVF) dataset indicating the success or failure of fertilization depending on the attributes like Age, Anti-Mullerian hormone (AMH), Right ovary (RO), Left Ovary (LO), Number of eggs, No of Inseminations, No of fertilized and Egg quality. This article mainly makes focus on variations of different ranges of cost ratio ? and establishes the possibility of reducing errors in the predictions made.