This is a script for predicting President Obama's approval rating. This was an early project of mine to learn about machine learning algorithms. I selected it in part because polling has such a rich dataset. Multiple pollsters poll the president's approval rating on a weekly or even daily basis.
The script uses a random forest regressor from Python's scikit-learn library. Data is collected from three sources: the Gallup and Rasmussen daily trackers as well as the Huffington Post poll collection. The script predicts the next poll for each source and provides a confidence estimate based on past results.