Leveraging ML Power for Crowdfunding Success Evaluation and Security Enforcement

Numerous online portals have been developed over time, to host large-scale and global crowdfunding campaigns
^[1]
in order to fulfill the funding goals of the listed projects. Therefore it becomes essential to be able to assess the probability of success of these campaigns based on the provided parameters. In this dynamic landscape of Online crowdfunding, ensuring the security
^[2]
and integrity
^[3]
of such a model to safeguard the interests of the investors is imperative to develop a high degree of credibility of the platform. This research introduces a sophisticated Machine Learning (ML) model
^[4]
which is tailored according to a specifically extracted dataset which works upon a comprehensive range of features like number of images, campaign timeline, textual description, security score, associated industry and other related attributes. A distinctive feature of this study includes engineering textual features
^[5]
from the campaign description to analyze its impact on the accuracy of the predictive model. One of the key components of this research initiative lies in discerning the scam or fraudulent campaigns listed upon the platform. The model was evaluated under several metrics to ensure its thoroughness. Notably, the Logistic Regression (LR) model was the best performer, showing a favorable ROC curve and a high accuracy of 0.899, which indicated that our predictive model was successful in its objective. Furthermore, by emphasizing the identification and suspension of fraudulent campaigns, our study fills a significant research gap and safeguards investor interests while promoting platform confidence. Our study increases investor trust in online crowdfunding ecosystems and by offering a strong foundation for well-informed decision-making. Our research offers a novel and thorough method for using machine learning to improve the effectiveness, security, and legitimacy of crowdfunding platforms.