Wondering how the prediction model works? Read about our methedology here.
We have open-sourced the prediction results of our machine learning model and some of the data we compiled for the model on Github. Check it out here. Feel free to download the data and reuse it in your own visualizations.
50Fifty uses Scikit-Learn for machine learning, running regression analysis on a variety of data points (called features), such as demographics and fundraising, from past elections. These classifiers then receive current candidates' features and compute their expected performance, in terms of both projected vote share and projected win probability, in the upcoming midterm elections.