Exemplifying the Effects of Distance Metrics on Clustering Techniques: F-measure, Accuracy and Efficiency
Tasleem Nizam, Syed Imtiyaz Hassan · 2020
Clustering is a type of unsupervised learning in which similar data or objects are collected into a single group. These objects can be exhibited in an n-dimensional Euclidean space. Hence based on distance metrics, similarities/dissimilarities can be measured. This paper reviews and analyzes two different clustering techniques' performance: K-means and FCM. These algorithms are applied on a dataset which contains information of those patients who had gone under breast cancer surgery. When K-means and FCM are applied using different distance metrics, it's found that precision, efficiency and accuracy of classification is also affected. On analysis, the K-means algorithm using Manhattan is found to be more suitable for clustering whereas FCM gives better result when Euclidean distance is chosen.