Optimized Classification for Organizational Workload
Orawat Yodnual, Roungsan Chaisricharoen · 2021
The classification of workload for companies or organizations consumes a lot of resources. Especially for smaller organizations, excessive resource use can lead to an imbalance in systems. Machine learning principles such as Naive Bayes can effectively reduce costs and manageability as it can reuse the organization's sources or statistics and can be reused without any administration. However, the performance of such an algorithm may deteriorate by itself due to improper deployment during data analysis. This study aims to optimize the Naive Bayes algorithm for classifying workload levels focusing on the distributions and the relevant kernels, including Gaussian, Kernel, Box, Epanechnikov, and Triangle. The optimized classification method offers better accuracy along with other performances. The results of the optimized method can be well applied to models that are similar to the proposed model.