크라우드펀딩 성공 예측 모델에 대한 다시점 분석: 기계학습 기법 활용

Considering financial risks for crowdfunding, it is important to predict the successful crowdfunding projects before the end of fundraising. Most of previous research on predictive models for crowdfunding success mainly analyzed crowdfunding projects which had ended fundraising. Therefore, there is a problem that creators and backers do not consider success timing of the projects. Examining crowdfunding projects in multi-phases, this research aims to develop a predictive model for successful crowdfunding projects in three stages. For this study, we collected 21,641 crowdfunding projects from Kickstarter, a representative crowdfunding platform, which were in fundraising process from August 15, 2016 to February 15, 2018. Then we applied several machine learning methods. Considering multi-phases in crowdfunding, we developed three stage prediction models and found difference from existing approaches. This study contributes to making three prediction models for success of crowdfunding projects in a different perspective and providing practical implications to both creators and backers of crowdfunding projects.