Quantitative Analysis of Gradient Descent Algorithm using scaling methods for improving the prediction process based on Artificial Neural Network

The health development is one of the most important challenges in the world today. All human beings are affected by many diseases due to various circumstances like pollution, climate change, living habits, etc. Therefore, the improvement of predicting diseases is a very essential process in medical management. Prediction refers to the results of an algorithm after it has been trained on a dataset. It is a mathematical process that seeks to predict future outcomes by analyzing methods. Classification methods of machine learning can be used to find accurate prediction of disease by the symptoms. This paper reviews the gradient descent algorithm such as Logistic Regression and Artificial Neural Network. These models are highly applicable and deliver reliable prediction accuracy with the help of a dataset. The survey indicates that the most popular classification techniques are Artificial Neural Network and Logistic Regression. The major purpose of this study is to investigate the performance of various scaling methods, including ensemble normalization and standardization methods, for improving disease prediction. The study also presents a performance comparison of classification algorithms, with and without applying the feature scaling of the data preprocessing techniques. In the proposed system, two algorithms, Artificial Neural Network and Logistic Regression, were used for the classification. Firstly, the accuracy of Artificial Neural Network and Logistic Regression without scaling method was calculated. The results show that Artificial Neural Network produces the highest accuracy of 86.13% in predicting heart disease. Next, various scaling methods were applied with Artificial Neural Network and Logistic Regression algorithms to improve the accuracy of the prediction process. The experimental results show that the accuracy of Artificial Neural Network using Ensemble Normalization and Standardization is 98.81%, which is greater than the other accuracies. Finally, a statistical test was used to assess the significance of the difference in performance among the classifiers.

Automated summary

The improvement of predicting diseases is crucial in medical management, as all humans are affected by diseases due to various factors. This paper reviews the gradient descent algorithm, including Logistic Regression and Artificial Neural Network, which are highly applicable and deliver reliable prediction accuracy. The study aims to investigate the performance of different scaling methods and classification algorithms for disease prediction.