Machine learning startup success prediction model

Abstract

Predicting the success of early-stage startups is a complex challenge with major implications for investment and entrepreneurial strategy. This study developed a machine learning-based prediction model using a dataset of 765 startups from Kenya and global sources. Key predictors of success included funding amount, age of company, and number of investors. Five models Logistic Regression, SVM, Random Forest, K-Nearest Neighbors (KNN), and Gradient Boosting were evaluated using accuracy, precision, recall and F1 score. The Gradient Boosting model achieved the highest performance of 98% training accuracy and 96% test accuracy, with an F1-score of 96% and AUC-ROC of 97%, making it the most effective predictor. A user-friendly interface was built to allow users to input data and receive real-time predictions along with strategic recommendations. The system serves as a practical tool for entrepreneurs and investors, enabling data-driven decisions based on key success factors.