A Predictive classifier for crowd funding success prediction using project attributes

Abstract

It won’t be an overstatement to say that, an internet service has become a commodity, a part and requirement of every household. This omnipresence of the internet has brought about major shifts in our lifelong ways, that is, we have a total new way of socializing; a disruptive educational model; paperless workspaces; branchless banks, just to name a few. All these innovations have been made possible with timely monetary support. In the meantime, a lot of good ideas have been gone down for nothing, just due to lack of social and monetary support. Among these technological contemporaries, there is also a total new way to get started with a business idea, which is crowdfunding. Crowdfunding is a new way, for funding projects, by raising money from a large number of random people. Given that, you would not be asking for just some interaction, but monetary contributions from people you haven’t met, and might possibly never ever be able meet them in person, easier said than done. All this puts focus on the launch of the project i.e. asking the right amount of funding for a particular type of project. So, here we have implemented and trained a crowdfunding project success predictor, such that, submitters, can improve their business pitch, before it gets too late. To improve the overall accuracy of the models, we scraped a huge Kickstarter dataset, and used it to train various models, to predict project success.