Executing on its mission of lowering the barriers to leadership, NationBuilder promises political candidates free access to its voter file, or an aggregate of the entire country’s voter registration data, including voting history. Customers can then leverage this data to develop and run machine learning models on how they should prioritize outreach activities. However, NationBuilder knew it had the opportunity to provide this data much more efficiently.
- Voter data is messy. NationBuilder needed to compile poorly-formatted and inconsistent datasets from hundreds of different state and county offices into a normalized nationwide voter file to perform machine learning.
- Custom data transformation tools are complex to build and difficult to maintain. Frequently changing data conventions at the state, county and city levels required NationBuilder to devote engineering resources to the constant maintenance of brittle custom tools.
- Voter datasets are large and constantly being updated. NationBuilder needs to refresh millions of voter records on a regular basis. Data refreshes must be fast and scale to support hundreds of millions of records.
Solution with Trifacta
- Trifacta Wrangler Edge deployed through the Amazon Web Services marketplace has dramatically reduced the time NationBuilder spends normalizing data to build a national voter file to develop and run machine learning models predicting election outcomes.
- NationBuilder has built a repository of Trifacta wrangle scripts that make it easy to refresh the file quickly whenever new data becomes available or timely updates are needed.
- By using Trifacta, NationBuilder has eliminated their need to use custom data transformation tools and expand upon their ability to perform machine learning against the national voter file.
NationBuilder is the world’s first software platform for leaders with one smart system to grow communities and lead people to action.