Validation on selected breast cancer drugs of physicochemical features by using machine learning models

Abstract

Breast cancer is one of the leading causes of death among females today. The elbow approach determines the ideal number of clusters after determining that the Dataset is highly cluster able with the Hopkins statistic. Three distinct groups with distinct differences were produced using the dataset's proposed expectation maximization fuzzy k-means clustering algorithm (PEMFKM). Different fuzzy clustering techniques, such as fuzzy k-means (FKM), fuzzy k-means with entropy (FKM.ENT), fuzzy k-means with entropy and noise (FKM.ENT.NOISE), Gustafson and Kessel - like fuzzy k-means (FKM.GK), Gustafson and Kessel - like fuzzy k-means with entropy regularization (FKM.GK.ENT), Gustafson and Kessel - like fuzzy kmeans with entropy regularization and noise (FKM.GK.ENT.NOISE), and PEMFKM, are evaluated. The partition coefficient (PC), partition entropy (PE), and Modified partition coefficient index (MPC) index values are better for FKM.GK than the suggested PEMFKM method. When compared to the FKM.GK method, the index values for the proposed PEMFKM algorithm have superior results for the parameters Silhouette (SIL), Xie and Beni index (XB), and fuzzy silhouette index (SIL.F). The results shows that the PEMFKM algorithm will provide better clusters and that the drugs in a given cluster may be combined for use in combination therapy for breast cancer treatment.