One of our first posts spoke at a high level of the technical problem CitizenNet is trying to solve. We are trying to predict which users matching a combination of demographic and interest targets will be interested in a post.
On the CitizenNet platform, a user creates a project that broadly defines the target audience, the pieces of Facebook content they are looking to promote, and other campaign and financial information.
Behind the scenes, a robust prediction system builds the targets for the project. This prediction system is trained on past behavior, which it then uses to predict how a future, unknown, project may be best targeted. One of the components of the prediction system is a classifier, which is a currently an ensemble of both Neural Networks and Random Forest classifiers. This blog post illustrates the basics of these methods.