Higher education institutions (HEI) generate a lot of raw data related to students, departments, personnel, courses, etc. “This information plays a fundamental role in decision-making. (…) A strategic viewpoint based on data optimization can show the true value of this information when it comes to making a university more innovative and competitive.”
The expected impact of Data Science on HEIs: 1) demographic changes will result in a decreased number of students; 2) cuts of budget and new financing models will lead to new ways of resources allocation and maintaining the teaching quality. Wikibon predicts the global market Big Data’s profits grow from 42 bill $ in 2018 to 103 bill $ in 2027. Statista foresees the grow of demand for Big Data management software from 14 bill $ in 2018 to 46 bill $ in 2027. The University Innovation Alliance in the USA analyzed data from 30,000 students of Georgia State to build a model with over 800 variables to identify students with the highest probability of dropping out. As a result, Georgia State University save students 15 mill $ in tuition in 2016 by adjusting the curriculum. “For every percentage point increase in student retention, Georgia State makes about USD 3 million more return on investment (ROI).” Arizona State University analyzed large raw data to help students understand mathematics. “The success rate went from approximately 65% to 85%.” Practices for efficient data management in HEIs: 1) structure the data by defining the set of operations it may be used for; 2) place connected data in a storage; 3) build and automate connections between the storage and analysis platforms to simplify data analysis; 4) introduce dashboards to have the data visualized for a quick overview of the organization’s situation. It is also important to provide a cultural change within a HIE and invest in data governance. Benefits for HIEs: a) problems are addresses specifically and with agility; b) ability to make improvements and predict scenarios; c) decision-making is based on reliable and real data.