Predictive Analysis of Crowdfunding Projects

Crowdfunding has become the social media version of fundraising campaigns whose underlying principle is to raise funds for a project from multiple people and to collectively accrue the required resources to make the project successful. The principal aim of crowdfunding platforms is to introduce budding entrepreneurs to an expanded pool of investors rather than traditional financial investors. Kickstarter is the largest reward-based crowdfunding platform which has successfully funded more than 2,00,000 projects and raised more than $6 billion. However, scarcely one-third of the projects are successful in reaching the funding goal before the deadline. Hence, reckoning the probability of success of a project is an interesting challenge. The proposed system helps classify a project as a success or failure. Supervised Machine Learning Models are implemented from which Random Forest provides the highest accuracy score of 90%. Regression Algorithms are implemented to estimate the funding a project is capable of achieving. Furthermore, BERT, spaCy, and TF-IDF are implemented to find keywords that affect the success of the project.