As academics, we all know the seriousness of paper retractions. Retractions indicate seriously flawed and unreliable research, errors, fraud, ethical issues, or other serious concerns. But can retractions be predicted? In this project, we will use data from "Retraction Watch" (http://retractiondatabase.org/), a database of academic papers that have been previously retracted, to investigate and develop methods for predicting whether a paper will be retracted or not in the future. An end-goal target is to build a binary classifier for "will be retracted" vs. "will not be retracted". A stretch goal would be to analyze any systemic biases in the classifier.