Educational data pre-processing from a domain-specific approach. Pre-procesamiento de datos educativos desde un enfoque de dominio específico.

Abstract

The data analysis processes to discover knowledge require pre-processing before applying techniques or algorithms to increase the quality of the data and adapt them to the formats that are best suited for processing, especially when the data comes from different sources. This article presents the experience in designing and constructing a strategy with a specific domain approach for the educational data preparation process. The study methodology included three stages: (1) design and construction of the strategy, (2) recognition and data selection, and (3) application of the strategy and review of results. The study was made up of data from the primary and secondary education system in the Norte de Santander department (Colombia). In addition, there was data referring to the enrollment process, which includes socioeconomic and family variables and data from evaluations of students' academic performance from three public educational institutions. For the two sources, data were processed from 2014 to 2018, with more than eight hundred thousand records. This work adds value in three main aspects: the scope concerning the educational level where the case study data comes from, the inclusion of the domain-specific approach in the solution, and the centralization of the data from multiple sources, resulting in data available in subsequent analysis processes. In conclusion, this work contributed both in the research field and applying knowledge in an existing case. Furthermore, it opened the possibility of carrying out subsequent tests with other data types from the educational context.