Predicting the performance of MSMEs: a hybrid DEA-machine learning approach

Abstract Micro, small and medium enterprises (MSMEs) dominate the business landscape create more than half of employment worldwide. How we can apply big data analytical tools such as machine learning to examine performance MSMEs has become an important question provide quicker results recommend better reliable solutions that improve performance. This paper proposes a novel method for estimating common set weights (CSW) based on regression analysis envelopment (DEA) operational research technique, which (i) allows measurement evaluations ranking comparisons MSMEs, (ii) helps overcome time-consuming non-convexity issues other CSW DEA methodologies. Our hybrid approach used several econometric techniques (such Tobit, least absolute shrinkage selection operator, Random Forest regression) empirically explain predict 5400 Vietnamese (2010‒2016), showed are efficient accurate ones. study, therefore, sheds new light two-stage literature, especially in terms predicting era strengthen role analytics management.