The professor posted one of the multiple competitions that we can work on for the project. Personally, I am pretty interested in the topic, but I will probably wait for a more financial oriented project to be released. This summary is the summary we have been provided:
Predicting the success of a project/campaign in its early stage is extremely important for crowdfunding environments (e.g., kickstarter). In this competition, you are challenged to build a predictor that can accurately predict the success chance of a kickstarter project after observing its early trend.
In particular, you have access to the following information:
- The requested money of the project
- The duration of fund raising campaign
You also observe for each project, during each of the first 10 days since its launch:
- The number of tweets that mention this project (containing a link to the project webpage)
- The number of replies to these tweets
- The number of retweets of these tweets
- The amount of pledged money
- The number of backers
which captures the social influence of the project during its early stage.
Ultimately, we have to create a classifier that will predict a model with high accuracy. The more accurate, the higher chance there is that you can make the top 10% and get bonus points.