Application of data analytics approach to spatial visualization test results

Abstract

The field of data analytics has received substantial attention in the past years due to global trend of collecting and analyzing data. Most of the attention and applications relate to consumers behavior, but the applicability of data analytics has extended to processes and market analysis. Data analytics can be considered a generic term used to refer to a set of quantitative and qualitative approaches that are applied to provide the basis for some decision making. The particular objective that is being pursued can be increase in productivity, additional business profit, or expected performance or behavior. Spatial visualization skills is something that has been linked to abilities to do engineering and technology work. There are several studies that have provided a relationship between the spatial visualization skills of students and their performance in engineering courses, particularly for engineering graphics and design courses. Similarly, there are reports that indicate the value in improving visualization skills when looking at the performance in learning in engineering courses, specifically for female students. This study is based on the application of data analytics approaches to spatial visualization scores with the goal of obtaining some predictive factors. The data utilized in this study is from the Purdue Spatial Visualization Tests with rotations (PSVT:R), which was administered to groups of first year students, all of them taking a course in engineering graphics. In addition to the scores for the test, demographic data was collected from the students as well as some background curriculum information, and such parameters are used in the data analytics approaches applied. The objective of the study is not to prove a specific trend or hypothesis, but to obtain results from the predictive analytic approach followed, so that, specific behaviors are identified and specific interventions are defined to address any specific behavior/factor. The software used in this study is RapidMiner, and different subsets of data are utilized in the machine learning phase, thus reaching more robust predictive conclusions.