Abstract:

This paper explores the use of educational analytic techniques to improve teacher training and governance in a large public school system in a developing country. Educational analytics can be performed at three different levels called Micro-, Meso-, and Macro-levels. Micro-levels analytics refer to analysis at the individual or student level. Meso-level analysis, on the other hand, is conducted at school or cluster level to improve the corresponding business processes. Macro-level represents highest granularity of analytics applied at the state, province, or country level. This paper shows how association rule learning can be used to improve teacher management in a developing country. Association rule learning is a method to discover interesting patterns of variables in large databases. The data used for this analysis is longitudinal data collected from approximately 1,400 primary public schools for 5 subjects from two districts of a province in developing country. This data has both governance and quality indicators. This educational data is collected each month by government officials on a regular basis. Results of applying association rule learning to this data can be used to guide decision-making for key stakeholders including donor agencies, district and department level of education administrators, teacher educators and school cluster heads operating at different levels of province, district, and school cluster. This paper first presents a profile this data according to its granularity, reliability, normality, and regularity and then makes recommendations/substitutions for missing data and identifies key variables of interest to the various stakeholders. A staged technique using cluster analysis in conjunction with association rule learning is then used to derive interesting patterns to enable decision-making at school and cluster level. This analysis guides in pinpointing professional development activities for specific groups of teachers. Reasons for low student score, consistently poor performance of students in a particular subject etc. can also be determined and necessary remedial actions can be taken.

Keywords:

Educational Analytics, Developing country, Teacher Management.