There is a political forecasting website called fivethirtyeight.com that has been famous since it correctly predicted the winning result for President Obama in 2008, with various sources of data collected, calculated and visualized in an easy-understanding way. However, the forecasting result for the 2016 election was not near correct, as you can see here(https://projects.fivethirtyeight.com/2016-election-forecast/?ex_cid=rrpromo). Because I have been interested in the power of statistical polling, and I have friends working as data scientists, they were quite surprised about the deviation from the prediction to the reality. For this course, let’s leave the statistical part aside, and talk about the visualized data.
We have two main candidates, Hilary Clinton for Democrats, and Donald Trump for the Republic, the supporters for them are displayed in blue and red respectively, and from the depth of color, we can decide whether a state has more Democrats or Republic supporters, and if most supporters are for one party. Those colors can also be used to present the candidates’ popularity over time, the swing probability for each state, the importance for the candidates to win a particular state, and so on. This gives me some insights on how to build a meaningful data visualization. Given a dataset, I need to firstly find out the parameters and their relationships with each other, what the dataset is about and how to display it. I can do it with time, do it with geography, with histogram or bubbles, maybe add some animations or not. The website fivethirtyeight.com can be useful for beginners of data visualization, to get started by learning and imitating. In the end, it’s not just easy-understanding that matters, it’s the way to present authentic data that will make the visualization good.