Mining Large Scale Data from National Educational Achievement Tests A Case Study
Reihaneh Rabbany, Osmar R Zaiane, Samira ElAtia
Abstract
Large scale analysis of educational assessment data, outlines patterns of success and failure, highlights factors of success, and enables performance prediction and eventually leads to proper ways of intervention. It has applications in both traditional settings where data is extracted from paper tests and surveys, and in e-learning settings such as distance, hybrid learning, and online courses. In the latter, drop out prediction and nding its factors and patterns is gaining much attention within the research community. In the former, the performance prediction is at the center of focus as drop outs are rare. Although the platform and data extraction is different, the essence of analyzing the test data is similar in both settings. In this paper, we present a case study on using data mining techniques in the analysis of large scale assessment data. The data is from the PanCanadian Assessment Program (PCAP), which is a national achievement tests administered by the Council of Ministers of Education, Canada (CMEC). The original ndings published based on this data underwent rigorous traditional statistical analyses. Here, we show new insights that could be obtained from the same data, by leveraging the power of Data mining.