Application of Bayesian Regularization Algorithm for Evaluation of Performance Software Complexity Prediction Model Based On Requirement

Abstract

Model performance evaluation is a method and process of evaluating the model that has been built. The model that will be evaluated is software complexity prediction model based on requirement. This model allows measuring software complexity before actual design and implementation. The experiment used three datasets namely training dataset, validation data set , and test dataset. For performance evaluation using Mean squared error. Mean squared error is very good at giving a description of how consistent the model is built. By minimizing the value of mean squared error, it means minimizing model variants. Models that have small variants are able to give relatively more consistent results for all input data compared to models with large variants. This research proposes the application of the Bayesian regularization algorithm for evaluating the performance of software complexity prediction model based on requirement. With this research it is expected to know how much the performance of the model that has been built.