Improved Fuzzy C-Means Clustering with Application for Patients of Iron Deficiency Anemia Discrimination | ||||
| The Egyptian Statistical Journal | ||||
| Article 1, Volume 49, Issue 2, December 2005, Page 93-116 | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/esju.2005.313558 | ||||
 View on SCiNiTO | ||||
| Abstract | ||||
| The Fuzzy C-Means Algorithms (FCMA) was applied for the discrimination between groups with Iron deficiency Anemia by artificial vision platelet counts of patient with certain measure. The aim of this study is to cluster, by FCMA, the measured learning and test data into 3 groups. The FCMA process was improved by the introduction of non-random initialisation of the cluster centers. The improved FCMA is a faster algorithm than the standard FCMA, it enables the investigator to test different feature vector representations quickly, which otherwise would have been impractical. In additions the Mahalanobis distance was used, instead of Euclidean for measuring the proximity of a pattern to a cluster distance. Furthermore, a classification approach with a reject threshold was investigated for increasing the classification performances. This was achieved by assigning the patients which were lying in the fuzzy boundaries between the available classes to reject class. The initialisation method was introduced in terms of the computation time and the percentage of correct recognition, a comparison study between Mahalanobis distance and Euclidean was carried for measure the error of classification. It was shown that the use of the Mahalanobis distance improved the performance in comparison to the Euclidean distance, the Mahalanobis distance allowed 95.7%, 95% of the learning and test sets to be correctly identified and also allowed an improvement in the correct recognition rates. A further experiment showed that the value of the stopping criterion & had little influence on the recognition rates, but it had a large influence on the computation time of FCMA. | ||||
| Keywords | ||||
| Fuzzy C-Means Clustering - Iron Deficiency Anemia; FCMA | ||||
|
Statistics Article View: 37 |
||||
