Table 1 Similar approaches for diabetic prediction using PIMA dataset

Kandha Swamy et al. [36]

Multiple ML based algorithms:

SVM, K-NN, J48 and Random Forest

73.82% with J48 classifier and claimed 100%

with K-NN

There is no adequate explanation is provided for the

pre-processing procedure  that was performed on the

dataset.

Yuvraj et al. [37]

Random Forest, Decision Tree and

Naïve Bayes classifier with data processing

Claimed 94% and 84% accuracies with Random

Forest Classifier and Decision Tree

Not specified how the data was pre-processed, although

 they did outline the Information Gain approach for feature

selection, which was utilized to extract the important features.

Sisodia et al. [38]

Decision Tree, Naïve Bayes and

SVM approach with Data Pre-processing.

Reported  highest accuracy of 76.30% with

 Naïve Bayes

Experimentation was carried out with 10 fold cross-validation,

and there was no more clear information on data processing.

Olaniyi et al. [39]

Multi Layer Feed Forward Network

(MLP-NN)

Reported 82% accuracy with MLP-NN

Before processing the data for classification, the authors

normalized the dataset in order to get a stable numerical

representation.

Ashiquzzaman et al. [33]

Deep Neural Networks with MLP, GRNN,

and RBF

Claimed an accuracy of 88.41%

The authors made a conscious decision not to pre-process

the dataset because DNN is capable of filtering the data

and acquiring the biases.

Zhou et al. [40]

Enhanced Deep Neural Network

Reported an accuracy of 94.02%

Model is primarily designed with the help of a deep neural

network’s hidden layers and it make use of dropout regula-

-rization in order to avoid over-fitting.

Yahyaoui et al. [41]

Convolutional Neural Network

Reported an accuracy of 76.81%

TThere is no adequate information on methodology

and techniques.

Naz et al. [42]

Decision Tree and Naive Bayes

96.62% and 76.33 % Accuracies reported

The authors worked on different classifiers

and reported accuracies in the range between 76% to 97%.

Abdulhadi et al. [43]

Random Forest Classifier

Reported an accuracy of 82%

TThere is no adequate information on data pre-processing

and methods.

Abdollahi et al. [44]

Ada boost algorithm

Reported an accuracy of 92%

This study aimed for integraion of different data mining

techniques and developed ensemble based training to

improve the performance.