testing a new forecasting model with the data from the 2019 #Argentina election. these are the results after processing polling data and estimating latent trends with a Bayesian algorithm and Markov chains. pic.twitter.com/2JUUEvq2GE— kenneth bunker (@kennethbunker) July 26, 2019
details of the forecast for the incumbent: Mauricio Macri pic.twitter.com/NAtVfqo5ip— kenneth bunker (@kennethbunker) July 26, 2019
Pollster Ratings: I took the average error of the difference between predictions and results of the 24 pollsters that fielded polls in the 2003, 2007, 2011 and 2015 elections. This is the summary (A=best; D=worst). I give a higher weight to polls from more accurate pollsters. pic.twitter.com/7PtcqSK4vs— kenneth bunker (@kennethbunker) July 31, 2019
How accurate were the polls in the past three #Argentina elections? Below you can see three graphs that show that as the election got closer (days) the polls got more precise. However, they still erred significantly, especially in 2007. The method did not make a difference. pic.twitter.com/puksCfMjKv— kenneth bunker (@kennethbunker) July 31, 2019
Relacionados
Los temas de campaña novedosos y de impacto los instala la oposición. El oficialismo reitera una serie de lugares comunes ya rituales - pero opacados por 4 años de pésima gestión - que se simbolizan en el mantra que repite la propaladora macrista hace un lustro: "se robaron todo"— Artemio López (@Lupo55) July 31, 2019
— Prol. Humor Gráfico (@HumorProl) July 30, 2019
1 comentario:
quiero creer que esos resultados para macri son exagerados
Publicar un comentario