A short post by me today. I wanted to look at the which states are important in winning the national election. Looking at the last 14 presidential elections, I generated the following correlation plot:
For those not familiar with how correlation plots work, the number bar on the right-hand-side of the graph indicates the correlation between a state on the left side with a state at the top, with the last row and column respectively indicating the national presidential election winner. Dark blue circles representing a correlation close to 1, indicate a strong relationship between the two variables, while orange-to-red circles representing a correlation close to -1 indicate a strong anti-correlation between the two variables, while almost white circles indicate no correlation between the two variables.
For example, one can see there is a very strong correlation between who wins Nevada and the winner of the national election. Indeed, Nevada has picked the last 13 of 14 U.S. Presidents. Darker blue circles indicate a strong correlation, while lighter orange-red circles indicate a weak correlation. This also shows the correlation between winning states. For example, from the plot above, candidates who win Alabama have a good chance of winning Mississippi or Wyoming, but virtually no chance of winning California.
This could serve as a potential guide in determining which states are extremely important to win during the election season!
Tomorrow is the date of the Canadian Federal Elections. Here are my predictions for the outcome:
That is, I predict the Liberals will win, with the NDP trailing very far behind either party.
It has certainly become the talk of the town with some of the latest polls showing that Donald Trump is leading Hillary Clinton in a hypothetical 2016 matchup.
I decided to run my polling algorithm to simulate 100,000 election matchups between Clinton and Trump. I calibrated my model using a variety of data sources.
These were the results:
Based on these simulations, I conclude that:
I think in the era of the 24-hour news cycle, too much is made of one poll.
A great deal of noise has been made in the previous weeks about the surge in the polls of Donald Trump and Bernie Sanders. This has led some people to question whether Hillary Clinton will actually end up being the Democratic party nominee in 2016. This was further evidenced by the fact that Sanders is now leading Clinton in the latest New Hampshire polls.
However, running an analysis on current polling data, I still believe that even though it is very early, Hillary Clinton still has the best chance of being the Democratic party nominee. In fact, running some algorithms against the current data, I found that:
Hillary Clinton: chance of winning Democratic nomination.
Bernie Sanders: chance of winning Democratic nomination.
These numbers were deduced from an algorithm that used non-parametric methods to obtain the following probability density functions.
Thanks to Hargun Singh Kohli for data compilation and research.