Abstract
This study investigates the impact of business environment obstacles on the performance of Cottage, Micro, Small, and Medium Enterprises (CMSMEs) in Bangladesh, utilizing data from 998 firms in the 2022 World Bank Enterprise Survey. A recursive feature elimination algorithm identified ten key business factors and obstacles from an initial set of twenty-eight that significantly influence CMSME performance. Analysis using ordinary least squares (OLS) and generalized least squares (GLS) regression models reveals that investments in electricity infrastructure, access to financial services, and obtaining quality certifications positively impact CMSME performance. In contrast, challenges such as power outages, delays in licensing, uncompetitive practices, and stringent tax and labor regulations hinder performance. Additionally, the predictive accuracy of the OLS model was compared with several machine learning algorithms, including decision tree, random forest, support vector, and gradient boosting, using a 75-25 training-testing split and k-fold cross-validation. The findings provide data driven actionable insights for policymakers to address specific obstacles, thereby enhancing the business environment for CMSMEs in Bangladesh.