Data Quality Indication for User Awareness and Automated Decision Making

Abstract

Enterprise Resource Planning (ERP) systems help organizations with administrating and planning various business related tasks and give insight with Key Performance Indicators. These mechanisms are highly dependent on being able to interpret the data in order to make the right decisions. This thesis defines a set of Data Quality Indicators (DQI) to calculate and visualize the quality of a large variety of spreadsheet data, used in ERP systems. The DQIs are used to complement a Machine Learning Classifier for automatic quality decision making. With a Support Vector Machine (SVM) approach, the system is able to correctly classify some spreadsheets. But data noise and some quality indicators not directly indicating real quality issues made it difficult for the SVM to clearly distinguish good spreadsheets from bad.