051

Title:

Assessing Students' Entrepreneurial Disposition by Text Classification Models

Discipline: Business Data Science

Presenter:

Yasser Farha

Abstract:

Researchers in Entrepreneurship Education measure student’s entrepreneurial orientation to assess the impact of external and internal factors, such as methods of teaching, cultural norms, and academic majors on that disposition. A variety of methods have been used for that purpose, and that includes using traditional survey-based instruments. However, the use of qualitative instruments is rare within the literature of Entrepreneurship Education impact studies. This lack of using qualitative instruments stems from the difficulties of analyzing such data. In addition, there is a reliability issue when the analysis is done by more than one person.

In our research, we introduce a text classification framework to classify new data that are collected through an open-ended survey, a qualitative instrument, into five classes. The classes stand for attributes that are linked to one of the four stages in Bygrave-Moore Entrepreneurial Process Model. We train models in the framework by using one thousand short text documents, as a training set, to predict classes of another three thousand documents, the testing set. The training set was manually coded by the authors. In both datasets, the students have been asked to answer two open-ended questions about motivations and discouragements of creating new ventures.

However, the framework proves its applicability and feasibility in analyzing new students’ data. The results are comparable with the outputs of manual coding and reach about 70% accuracy. The purpose of this framework is to automate the analytics process with which to enhance the efficacy of Entrepreneurship Education.

Author(s):

Yasser Farha, Cesar Bandera

Funding Acknowledgements:

SACM