In United States, there are 13 appellate courts that sit below the U.S. Supreme Court. All 94 federal judicial districts are organized into 12 regional circuits, each of which has a court of appeals. The appellate court’s task is to determine whether or not the law was applied correctly in the trial court. Appeals courts consist of three judges and do not use a jury.

In this machine learning project, our major goal is using data from the past 134 years of circuit court appeals hearings to predict whether Courts of Appeals would affirm or reverse the trial court decisions, along with analyzing the causal factors influencing the judges’ decisions. Potential factors include political parties the judges belong to, which indirectly affect personal sentiments recorded in the texts. Further and deeper conclusion can be made from our results, for example, whether this case could help legislation toward a certain direction and influences from the past cases.