Abe Eyman Casey, Alex Son, Brodie Thiede, Jack Welsh
Predictions
We predict 2414187 votes or 54.7% voter turnout for all of Minnesota
Most Important Predictors
Methods
Twitter API
Used Twitter’s APIs to gather Tweets from the profiles of candidates from MN running for office during the 2012, 2014, 2016 and 2018 elections
Collected metrics and utilized natural language processing and sentiment analysis on the Tweets’ content
Random Forest
We choose to use Random Forest (the randomForest package in R) to investigate turnout.
We chose Random Forest because of its overall robustness. Random Forest doesn’t have conditions and doesn't have problems with multicollinearity if given correlated explanatory variables.
Cross Validation - We performed 100 fold cross validation using the rfUtilities package in R to examine the performance of our models.
Data Sources
Where we got our data from
American Fact Finder for all demographic data
Minnesota Department of State for historic voter turnout data
Twitter API for all Twitter data.
OpenSecrets.org for the candidate fundraising data.
Minnesota State Demographic Center for the economic data
Predictions
We predict 54.7% voter turnout for all of Minnesota