A Robust System for the Evaluation of Innovation and Entrepreneurship Education using Grid Search with SVM

For the past few years, Innovation and entrepreneurship have become key factors in economic growth and development; but, still evaluation of innovation and entrepreneurship education and their effectiveness remains a critical challenge. Even though previous researchers suggested numerous methods that contain many indicators still research in this field is essential to accurately classify the effectiveness of innovation and entrepreneurship. In this study, key indicators are determined at an early stage that assist in evaluating innovations and entrepreneurship of various domains and further clustered using the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm to combine the indicator based on their specificity and respective domains. Then, Back Propagation (BP) is employed to regulate the effectiveness of each indicator based on its weight. Ultimately, grid search with a Support Vector Machine (SVM) classification model is imposed to categorize innovation and entrepreneurial projects and ideas effectively. From the results, this DBSCAN approach acquired an accuracy of 98.95 % outperforming conventional methods such as Random Forest and thus providing a robust solution to many policymakers to effectively evaluate innovations and foster entrepreneurial activities.