Track: Engineering Education

Abstract

This research compares models based on artificial neural networks (ANN) to predict entrepreneurial intentions among undergraduate students according to the results of the University Entrepreneurial Spirit Students’ Survey (GUESSS) of 2016. The research is carried out following a classic 4-stage methodology (analysis, design, development, and validation). During the analysis, surveys were thoroughly reviewed and preprocessed. During the design, the survey’s questions are combined according to certain criteria to build 10 classification models. Construction and validation are carried out entirely using the software WEKA. For the purposes is this investigation 627 surveys are considered. The dataset is split up in two subsets: 80% for training and test, and the remaining 20% for validation. The approach to predict entrepreneurial intentions considers building and comparing 10 ANNs. The results reveal that, with a heavily imbalanced dataset, the proposed models classify correctly between 77% and 80%. However, the area under the curve ROC present low values. In conclusion, the investigation results show that predictive models based on ANN can help predict the entrepreneurial intention of undergraduate students by means of knowing some information about their family background, social environment, and university. However, these results might not be conclusive since the dataset is significantly imbalanced.