A Novel Architecture for Diabetes Patients’ Prediction Using K-Means Clustering and SVM

Table 1

Comparative study of existing approaches used by the researchers and accuracy achieved.

Sr. no.	Method used	Accuracy (%)	Reference
1	Decision tree	78.17	[10]
2	Higher-order NN with PCS	89.47	[11]
3	NN	93.5	[12]
4	Classifier using the K-means algorithm with logistic regression	98	[13]
5	Fuzzy K-nearest neighbors	89.1	[14]
6	GA combined with multilayer perceptron neural network	79.13	[15]
7	Class-wise K-nearest neighbor (CkNN)	78.16	[16]
8	Multilayer feedforward neural network	95.5	[17]
9	F-score, K-means clustering along with Z-score normalization and SVM	98	[18]
10	Ant colony optimization (ACO)	84.24	[19]
11	Re-RX with J48 graft, combined with sampling selection techniques	83.83	[20]
12	Information gain (IG) along with deep NN	90.26	[21]
13	Decision tree and naïve Bayes	76.9 and 79.5 respectively	[22]
14	ANN and FNN	86.8	[23]
15	SVM with an RBF kernel and with a polynomial kernel	82.2	[24]
16	K-means clustering along with GA and CFS	96.68	[25]
17	K-means clustering combined with decision tree C4.5	93.33	[26]
18	GA and back propagation network (BPN)	77.7	[27]
19	General regression neural network (GRNN)	80.21	[28]
20	Random forest and gradient boosting classifiers	90	[29]
21	Covering-based rough set	79.34	[30]
22	SVM (with RBF kernel)	75.5	[31]
23	Amalgam KNN	97.4	[32]
24	K-means clustering combined with decision tree C4.5	92.38	[33]
25	Fuzzy C-means combined with SVM and KNN and weighting methods (FCMAW)	91.41 and 84.38, respectively	[34]
26	GDA and least square support vector	82.05	[35]
27	Random forest combined with recursive feature elimination	73	[36]
28	Neural network model with backward elimination feature selection method	84.52	[37]
29	RB-Bayes	72.9	[38]
30	Naïve Bayes	76.3	[39]
31	Deep neural network restricted Boltzmann machine	80.9	[40]
32	Goldberg’s GA combined with multi-objective evolutionary fuzzy classifier	83.04	[41]
33	Neural network and ANFIS structures	81.3	[42]
34	Cartesian genetic programming	80.5	[43]
35	Improved the K-means and the logistic regression	95.42	[44]
36	SVM combined with neural network	88.04	[45]
37	KNN	82.29	[46]