Recommend Missing Information System in the Existing Application using SMOTE and Multi-Layer Perceptron Neural Network

  • Sumitra Nuanmeesri

Abstract

The objective of this article is to develop the recommend the filling of missing data in the existing application by improving the dataset using the synthetic minority over-sampling technique (SMOTE) and creating predictive models in the neural network technique. The results of the simulation model’s effectiveness test using the 10-fold cross-validation technique with the SMOTE and the Multi-Layer Perceptron Neural Network technique provide the accuracy of 96.25% at the size of data set was 500% which is higher than the simulation model which applies the Multi-Layer Perceptron Neural Network technique alone. The overall model effectiveness evaluation from experts and users found that efficiency is the highest level and is acceptable with consensus. The system was deployed as web services on the existing application. It helps improve the efficiency of the user's operation by recommend the suitable values for the missing information and notify the message to the users.

Published
2020-11-05
How to Cite
Sumitra Nuanmeesri. (2020). Recommend Missing Information System in the Existing Application using SMOTE and Multi-Layer Perceptron Neural Network. International Journal of Advanced Science and Technology, 29(04), 10943–10954. Retrieved from http://sersc.org/journals/index.php/IJAST/article/view/33603