Site icon R-bloggers

Argentine’s 2015 Presidential Election Forecasts

[This article was first published on » R, and kindly contributed to R-bloggers]. (You can report issue about the content on this page here)
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.

Predictions

The model I built to forecast the Argentine’s 2015 presidential election indicates the official candidate have a significant chance of making it right way this Sunday, avoiding a runoff with Mauricio Macri in November. The electoral preference distributions are quite apart from each other, with the distribution of Daniel Scioli about 40% of the positive vote.

Supported by the outgoing president Cristina Fernandez, Daniel Scioli, emerged as front-runner at the turn of the year. The model built on more than 115 polls by 24 pollsters suggests he has 86% of probability of finishing above 40% of the positive votes, and more than 10% distante from Mauricio Macri, his nearest rival. To avoid a runoff, a candidate must obtain more than 45% of the valid votes, or more than 40% with a difference greater than 10% from the second candidate. The typical difference the model predicts is about 12%; not too loose, but a double digit difference.

(The models I run for the primaries)[http://danielmarcelino.com/r/08-2015/Argentine-general-election-2015/], predicted quite well the proportion of votes each major candidate received, if fact, the model with a Dirichlet trendline appreciated slightly the front-runner candidate but the loess model.

When did the Argentine presidential election approach its tipping point?

The first poll I’ve in my database dates back to March 2014. From this period, the only significant movement in the preference distribution happened before the primaries, in August, and this was true for both Scioli and Macri. Though for Macri, the upward movement was imperceptible. Because nothing change since then, it’s hard to say the candidate gained momentum,as his numbers have neither risen nor fallen ever since.

The model

The polls were modeled using a Kalman Filter, then probabilities were computed using a Dirichlet Multinomial Distribution. Details on the model can be found within (my gists)[https://gist.github.com/danielmarcelino].

To leave a comment for the author, please follow the link and comment on their blog: » R.

R-bloggers.com offers daily e-mail updates about R news and tutorials about learning R and many other topics. Click here if you're looking to post or find an R/data-science job.
Want to share your content on R-bloggers? click here if you have a blog, or here if you don't.