Understanding Political Performance and Improving Citizen Engagement through Data
Big data analytics are changing the way in which sports are played and understood, resulting in improved performance and increased fan engagement. Can a similar virtuous data cycle be applied to political games to increase understanding of political performance and improve citizen engagement?
Apply mathematical, statistical, computational, and visualization tools to political data to help improve fact-based political storytelling for mass media and social media contexts.
Python, data visualization tools such as https://d3js.org/ and PowerBI, some familiarity with cloud computing environment such as AWS, data analysis. https://www.unrealengine.com/en-US/unreal-engine-5 knowledge could be an asset.
Data & Metadata
Data & Metadata: RapidAPI – (data links to be provided) Datasets provided to support this problem are proprietary Goverlytics datasets extracted from Canadian and USA legislatures.
- Topic Modeling in Embedded Spaces https://arxiv.org/abs/1907.04907
- Previous M2PI projects
- RapidAPI – with keys to be provided soon
Principal Component Analysis (PCA), as well as Factor Analysis, are a couple of techniques used to increase data value by making data more interpretable while simultaneously preserving as much variability and information possible (https://royalsocietypublishing.org/doi/10.1098/rsta.2015.0202). Given large topic-indexed datasets reflecting activity by parliamentarians such as chamber interventions, committee interventions, bills, motions, and chamber votes how might such analytical techniques be used to reduce the dimensionality of these sets while increasing their interpretability ? Can useful and efficient graphical displays for the public be generated through the application of such techniques to political data? What other types of data analysis methods may be used alongside such techniques to extract meaning from political data? What measures of similitude or difference between individual politicians or parties might be derived? How might such features help to measure political performance? How can topic indexes be aggregated to reflect similarities in political concern ?
Data Storytelling Pro Tips
Five components are key to storytelling: characters, setting, plot, conflict, and resolution. These essential elements engage your audience. When telling stories with data, use a story framework: have data to describe characters (parliamentarians in our case), threats (poor policy?), goals (better policies?), changes (how are things may be evolving over time). https://online.hbs.edu/blog/post/data-storytelling.