Temporal Trends Analysis for Dengue Outbreak and Network Threats Severity Prediction Accuracy Improvement

Sharef, Nurfadhlina Mohd; Husin, Nor Azura; Kasmiran, Khairul Azhar; Ninggal, Mohd Izuan Hafez

doi:10.6025/jdim/2019/17/3/122-132

Cited by 1 publication

(1 citation statement)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They have applied the model to segment microscopic blood image using a neural network to improve results about dengue fever. [20][21][22][23][24][25][26][27] Back propagation network applied to classification result. It gives 98% accuracy.…”

Section: Literature Surveymentioning

confidence: 99%

Multi and Binary approaches on Dengue by Stochastic Learning

G.¹,

A.²

2023

INDJCSE

View full text Add to dashboard Cite

Dengue is an arboviral disease caused by the Aedes mosquito-borne dengue viruses. The huge volume of dengue cases ever reported globally and prompted WHO to declare the virus as the world's top 10 public health threats. This works governs that the stochastic learning approaches on dengue classifications. The Multi class learning model that utilises the stochastic gradient descent (SGD) algorithm yields the highest accuracy value, which is 91.63% accuracy. The binary class that uses the SGD learning model generates the accuracy value of 86.49%, which is the lowest value possible. The Multi class with SGD learning model yields the highest positive predictive value, which comes in at 0.92 for this metric. The binary class using the SGD learning model produces the lowest positive predictive value of 0.85. This is the result. The Multi class with SGD learning model generates the highest true positive rate value, which is 0.92 of the total possible value for the true positive rate. The binary class that uses the SGD learning model produces the value of 0.87 for the true positive rate, which is the lowest possible value. The AUCROC number that is produced by combining many classes with the SGD learning model is the highest possible, coming in at 0.93. The binary class using the SGD Text learning model produces the AUCROC value of 0.50, which is the lowest possible value. The AUCPRC value that is produced by the Multi class with SGD learning model is the highest, coming in at 0.93 of the total AUCPRC value. The binary class using the SGD Text learning model produces the AUCPRC value of 0.79, which is the lowest possible result. The F1-score value that is lowest belongs to the binary class, which uses the SGD and SGD Text learning models. When compared to other learning models, the Multi class with SGD Text learning model offers the greatest F1-Score value. The F1-score value that is obtained using the SGD learning model for the binary class is the lowest possible, coming in at 0.86. The value of Cohen's kappa statistic is 0.30 when applied to the binary class using the SGD learning pattern. When compared to other models, the Multi class employing the SGD Text learning model is demonstrating the highest level of relevancy, as measured by the K statistic value. The phi coefficient has a value of zero when applied to the binary class using the SGD Text model. The phi coefficient value that is displayed by Multi class with SGD Text learning pattern is 0.6, which is the greatest possible value for the phi coefficient. The phi coefficient value that is displayed by Multi class with SGD Text learning pattern is 0.6, which is the greatest possible value for the phi coefficient. When compared to other models, the SGD learning algorithm produces the best results with the least amount of variation for the dataset that was borrowed.

show abstract