Development of Business Intelligence System and Prediction with Data Mining of Lam Sam Kaeo Town Municipality, Thailand

Abstract

The purpose of this research was to develop a business intelligence system and make predictions with data mining of Lam Sam Kaeo Town Municipality. The process of development used Microsoft Visual Studio (SSDT) with SQL Server Integration Services (SSIS) to create a data warehouse, then using SQL Server Analysis Services (SSAS) to create cube and using SQL Server Reporting Services (SSRS) to create reports then publish to web browsers for local administration officers to make decisions. The sample data used in this project covered local government taxpayers living in Lam Sam Kaeo Town Municipality in 2020: altogether about 200 taxpayers. The analysis was based on the new land and building tax act 2019. RapidMiner Studio was used to create the analysis model to determine the factors that cause tax delays in Lam Sam Kaeo Town Municipality. A comparison of three classification algorithms showed similar accuracy: J48 Decision Tree has accuracy = 96%, Naïve Bayes has accuracy = 92%, Neural Network has accuracy = 96.50%, ANOVA test found no significant difference at 0.05 level is not different so the researcher chooses the Decision Tree method for this research. The results are that the most influential factors causing overdue tax payments are ‘forget to pay’, ‘lack of advertising’, and ‘lack of e-payment method.