Can It Screen? Exploring the Usability of a Data-driven Lean Canvas Framework for Startup Selection in Accelerators

The emergence of startups across various industries has resulted in an influx of new ventures seeking guidance from entrepreneurial programs to assist in their development, including startup accelerators. However, given the human capital and time constraints faced by accelerators, issues arise on selecting the most optimal startups. This thesis proposes a framework that would leverage startup’s application into an accelerator to create insightful features, through machine learningtechniques. Furthermore, the leancanvasframeworkwould be utilized to mapthe features to its respective dimensions and identify the impact each dimension has on the startup selection process. I extensively studied the effectiveness of this framework by analyzing startup’s application to a US-based accelerator. The most important features were a startup’s competitive advantage and industry similarity with accelerator programs. The proposed pipeline introduces a unique framework to assist in the startup selection process and contributes to the growth within the entrepreneurial ecosystem.