Determination of essential features for predicting start-up success: an empirical approach using machine learning

Abstract

Various significant features influence the success of a startup in an entrepreneurial ecosystem. However, identification of such features has not been performed coherently. This study aims to bridge the gap by determining the essential features impacting a startup's success. In the first part, a machine learning-based approach is applied empirically to determine critical features from ten datasets. Then the obtained features were validated using popular feature selection techniques based upon Recursive Feature Elimination (RFE), SelectKBest, and XGBoost to assure the final significant features. Furthermore, mutual information is used between the features obtained from the proposed approach and the ground truth to measure the approach's efficacy, which is calculated to be 90%. The significant features acquired through this approach imply how the entrepreneurs prioritise the determining features in startup success and maximise the socio-economic outputs. This study provides an essential contribution to the entrepreneurial finance literature by empirically exploring the features influencing a startup's success based on its financial performance in different stages of its lifecycle. Startups can apply the implications of this study to reduce uncertainty and enhance performance. It also helps founders, investors, and researchers understand a startup's performance and plan future steps accordingly.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 12 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.