A similarity measure of intuitionistic fuzzy soft sets and its application in medical diagnosis

• The proposed method is applied on UCI Machine Learning Repository datasets and its corresponding performance measures are obtained.
• The comparison of the proposed similarity measures with the existing results reveal that the accuracy obtained by the present technique provides better similarity rating.
• It is also observed that fixing the threshold values for the similarity measures is difficult due to the negative rating in some existing methodology.
• The above difficulty can be avoided by using this newly proposed technique, in which the similarity measures lies between 0 and 1.
• The similarity rating obtained by the suggested technique is compared with the class distribution data set and are analyzed.
• The proposed method exhibits more accuracy, sensitivity, ROC curves, AUC values and enhanced F-measure than the existing methods.

In this paper, a new similarity measure and a weighted similarity measure on intuitionistic fuzzy soft sets (IFSSs) are proposed and some of their basic properties are discussed. Using the proposed similarity measure, a relation (≈ α) between two IFSSs are defined and it is found that the defined relation is not an equivalence relation. Further, the effectiveness of the proposed similarity measure is demonstrated in a numerical example with the help of measure of performance and measure of error. Moreover, medical diagnosis problems have been exhibited through a hypothetical case study by using this proposed similarity measure. Finally, the proposed method is applied to 10 different medical data sets from UCI Machine Learning Repository datasets and its similarity measures are calculated. The corresponding performance measures, like, accuracy, sensitivity, specificity, ROC curves, AUC values, and F-measures are obtained and it is compared with the existing methods. This shows that the proposed method exhibits more accuracy, sensitivity and enhanced F-measures than the existing methods.

Figure optionsDownload as PowerPoint slide