Abstract

The increasing complexity of innovation ecosystems demands a quantitative understanding of how agility influences success in entrepreneurship and project management. This study proposes the Agile Readiness Index (ARI) as a composite, data-driven indicator for assessing team agility and adaptability in student-led startup projects. Using the Student Startup Success Dataset (n = 2,100, Kaggle, 2019–2023), which contains features such as innovation score, mentorship, incubation support, and funding, the research applied Principal Component Analysis (PCA) to extract a latent dimension representing agile capability. The first principal component, explaining 10.53 % of total variance, was normalized to construct the ARI and subsequently integrated into supervised learning models to predict startup success. Four machine-learning classifiers—Logistic Regression, Random Forest, XGBoost, and Support Vector Machine (SVM)—were trained and evaluated. Logistic Regression achieved the highest performance (accuracy = 98.1 %, AUC = 0.9995), followed by SVM (AUC = 0.9953), confirming that agile-related factors maintain a primarily linear relationship with success outcomes. Feature-importance and SHAP analyses identified funding amount, innovation score, and the Agile Readiness Index as the top three predictors, emphasizing the interplay between resource availability and adaptive learning capacity. Moreover, the interaction effects of team experience indicated that moderate experience levels, coupled with mentorship, enhance agile responsiveness, while extremes in experience reduce flexibility. The findings demonstrate that agility can be empirically modeled, interpreted, and predicted through computational analytics. The proposed Agile Readiness Index not only improves predictive accuracy but also strengthens theoretical understanding of how adaptive behavior and feedback integration shape entrepreneurial outcomes. This research contributes a replicable framework for quantifying agile maturity across educational and organizational innovation contexts, bridging the divide between agile management theory and evidence-based practice.

Keywords: Agile Readiness Index, Machine Learning, Principal Component Analysis, Startup Success Prediction, SHAP Explainability

How to Cite:

Widodo, S. & Kafilla, P. I., (2026) “Predicting Student Startup Success by Modeling an Agile Readiness Index with PCA and Machine Learning”, Agile Management 1(2), 75-91. doi: https://doi.org/10.63913/am.v1i2.128