I want to carry out a project about demographics , elections and income levels in the US.
Majority of the votes of 2016 general elections were related to income levels and/or unemployment rates of the states in the US.
Here, I am planning on calculating correlation between vote percentages and income levels for each state. As for visuals, a bar chart could convey the concept well.
Outcomes of 2016 general elections were linked to racial distributions in each state.
I have the racial distribution of every state of the US at hand. The votes that republicans and democrats got in percentages should match to that pattern. Or else the hypothesis will be rejected.
An interactive visual on US map using D3 may foster visualising the info in this case. A pie chart of regional totals i.e southern states, mid-west etc. may come in handy.
Unemployment rates and income levels in each state should be highly correlated in the US.
Here again I will display the patterns of unemployment and income levels state by state. The visual will be again a US map of where they match and where they don’t.
The other idea of mine to prove / dispove the very first hypothesis is to combine the first and third one and use an data mining program. I am going to choose some states i.e around 40, as test cases. Label the outcomes as either Democrat or Republican. Then, make the program guess the rest and check the accurracy, confusion matrix.
I may use a regression model for the second hypothesis as well. I highly doubt that relation will be linear. However, within some error the regression may give an idea.