Scholarlink

<p>Our contribution fits within the broader trajectory from traditional econometrics to predictive machine learning and advanced AI-driven decision support, focusing on a distinct niche in digital entrepreneurial finance: the application of interpretable machine learning to predict post-campaign outcomes of equity crowdfunded firms. Specifically, we combine prediction with explainability to identify which features are most associated with transitions to external post-campaign scenarios. This is enabled by a hand-collected dataset of 708 post-initial and seasoned campaign scenarios across multiple platforms, spanning 2012–2024, including high-dimensional and multi-category features. This yields more representative samples and more generalizable predictive insights. Firm valuation and campaign characteristics are strong predictors. Additional features, including firm age, core team size and female representation also matter beyond mechanical campaign-related outcomes. Moreover, the model’s predictions show that synergistic interactions between firm valuation and campaign indicators yield higher predicted probabilities than any single predictor alone, highlighting the importance of multidimensional interactions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Predicting post-campaign scenarios of equity crowdfunded firms: a machine learning approach with interaction effect analysis

  • 期刊论文

  • 发表日期：

    2026年7月3日

查看全文
  • Ines Gafrej,
  • Oussama Gafrej,
  • Houssam Bouzgarrou

摘要

Our contribution fits within the broader trajectory from traditional econometrics to predictive machine learning and advanced AI-driven decision support, focusing on a distinct niche in digital entrepreneurial finance: the application of interpretable machine learning to predict post-campaign outcomes of equity crowdfunded firms. Specifically, we combine prediction with explainability to identify which features are most associated with transitions to external post-campaign scenarios. This is enabled by a hand-collected dataset of 708 post-initial and seasoned campaign scenarios across multiple platforms, spanning 2012–2024, including high-dimensional and multi-category features. This yields more representative samples and more generalizable predictive insights. Firm valuation and campaign characteristics are strong predictors. Additional features, including firm age, core team size and female representation also matter beyond mechanical campaign-related outcomes. Moreover, the model’s predictions show that synergistic interactions between firm valuation and campaign indicators yield higher predicted probabilities than any single predictor alone, highlighting the importance of multidimensional interactions.