Experimental Evaluation of Autonomous Open Data Prediction Framework (AODPF)

Enterprise resource planning (ERP) systems need reliable data that can be verified to allow making various forecasts, thus increasing the company's added value. Therefore, there is a need to have different methods to make forecasts automated and find the best methods with specific data outside ERP systems. This paper represents the final results of the AODPF using road maintenance cases. The AODPF framework approaches automatically find the most suitable method using ARIMA, missing point filling, Kalman filter, and additional datasets in combination. The AODPF is presented as a framework solution. An experimental design with experimental scenarios and experimental factors has been developed within the framework of the study. The experiment results have been tested with t-test and ANOVA approaches. It has been confirmed that using a framework that combines various factors and sources of data helps improve forecasting accuracy outside ERP systems.