Predicting Chronic Kidney Disease Using Hybrid Machine Learning Based on Apache Spark

Table 13

The comparison of performance between the previous studies and our work on the same dataset.

REF	Feature selection methods	The best model	Dataset	Result
[22]	RFE	RF	CKD dataset	AC = 100%
				PR = 100%
				RE = 100%
				FS = 100%
[27]	No	A hybrid model LR and RF	CKD dataset	AC = 99.94%
				E = 99.84%
				S = 99.80%
[30]	CFS	AdaBoost based on KNN	CKD dataset	AC = 98.1%
				PR = 98%
				RE = 98%
				FS = 98%
[23]	Rffs, FS, FES, BS, BES	RF	CKD dataset	AC = 98.825%
				RE = 98.04%
[24]	Cost-sensitive ensemble feature ranking	An ensemble of decision tree models	CKD dataset	AC = 97.27%
				PRC = 99.44%
				RE = 96.25%
				FS = 97.68%
[25]	No	Random subspace-based KNN	CKD dataset	AC = 100%
				RE = 100%
[26]	Genetic search algorithm	Multilayer perceptron	CKD dataset	AC = 99.75%
Our work	Relief-F	DT	CKD dataset	Cross-validation result AC = 100%, PRC = 100%, RRE = 100% FS = 100% result of testing AC = 100%, PRC = 100%, RRE = 100%, FS = 100%
		GBT Classifier	CKD dataset	Cross-validation result AC = 100%, PRC = 100%, RRE = 100%, FS = 100%; result of testing AC = 100%, PRC = 100%, RRE = 100%, FS = 100%