EXPLORING MACRO-LEVEL EDUCATIONAL ANALYTICS TO IMPROVE PUBLIC SCHOOLS IN A DEVELOPING COUNTRY
Educational analytics can be conducted at three distinct levels; macro, meso and micro . The Micro-level analytics is performed for individual students or groups of students within a cohort or a school to identify students’ strengths, and weaknesses, and to predict their success. The Meso-level educational analysis operates at school and school cluster level to improve school’s performance and business processes. Macro-level analytics is the highest level of education analytics which is performed at a region or state level by analyzing data from many schools and school districts. Education data mining has been previously used to discover patterns in students’ learning, understanding learning behavior of a sample of students, and for modeling and analyzing students’ learning by means of online intelligent systems . However, a brief survey of literature shows that while a host of data mining techniques have been used, their scope has been mostly limited to the Micro or Meso-level analytics . This paper explores how data mining techniques can be used to enhance education in developing countries at the macro level. Such an analysis can potentially enable governing officials and policy makers to make informed decisions for continuously improving the educational system of a province or a country. The exploration reported in this paper is conducted on a rich set of data including independent learning outcome assessments of multiple subject areas and teacher training, economic and human development data. This data is collected on a monthly basis for every public school in one province of Pakistan. According to Education for All Global Monitoring Report Fact Sheet, Pakistan has the worst education indicators globally, as it stands second among the countries with children out of school, and ranks 113th out of 120 countries in the Education Development Index. A design for Macro-level educational analytics of about 15,000 schools will be presented. A host of data mining technique will be explored in this paper. For example, association rules can help identify the pertaining social and economic factors that result in the degradation of an entire educational system. Factors like economic background, gender of students and teachers, geographical area, facilities, and condition of the schools can be associated with each other, and with the quality of education. Similarly, use of clustering techniques can help answer potentially important questions like how many schools have similar problems, how many teachers have similar training needs, how many students could not focus on their studies because of their socio-economic problems, etc. Finally, classification techniques can be used to classify schools, and teachers according to their performance. The classification results can further be used to determine the factors that need to be considered and implemented for other schools and teachers to achieve a systematic improvement in the entire educational system.
 S. B. Shum, "Learning Analytics," UNESCO Policy Brief, 2012.
 R. S. Baker, "Educational Data Mining: An Advance for Intelligent Systems in Education," Intelligent Systems, IEEE, vol. 29, pp. 78-82, 2014.
 S. Borkar and K. Rajeswari, "Predicting Students Academic Performance Using Education Data Mining," International Journal of Computer Science and Mobile Computing, vol. 2, pp. 273-279, 2013.