Making Sense of Unstructured Data: An Experiential Learning Approach

The need for competent data scientists is recognised by industry practitioners worldwide. Currently tertiary education institutions focus on the teaching of concepts related to structured data (fixed format), for example in database management. However, the hidden value contained in unstructured data (no fixed format) motivated the need to introduce students to methods for working with these data sets. Therefore, an experiential learning approach was adopted to expose students to real-life unstructured data. Third year students were given an assignment whereby they could use any publicly available un-structured data set or an unstructured dataset supplied to them following a set methodology (CRISP-DM) to discover and describe the hidden meaning of the data. As part of the assignments students had to reflect on the process. Twenty student assignments were analysed in an attempt to identify the effectiveness of the experiential learning approach in the acquisition of skills pertaining to unstructured data. Our findings indicate that the experiential learning approach is successful in the teaching of the basic skills needed to work with unstructured data. We discuss the appropriateness of the prescribed methodology, the students' performance, and lessons learnt. On the basis of these lessons we conclude with some recommendations for educating future data scientists.