llambourne / errorbar-proportion Goto Github PK

Make a quick adaption of current private version that uses statsmodels.stats.proportion.proportion_confint with method='jeffreys' as default instead of the Bayesian method with uniform prior, since that seems to be just used in physics and the Jeffreys prior seems to be favoured by statisticians. Also, will try and use the bar method, instead of combining bar and errorbar since it seems to take errors now and there was some recent method for writing above the bars to try and use.

Confidence interval options are described here:
https://en.wikipedia.org/wiki/Binomial_proportion_confidence_interval
and here:
Brown, Lawrence D.; Cai, T. Tony; DasGupta, Anirban (2001). “Interval Estimation for a Binomial Proportion”, Statistical Science 16 (2): 101–133. doi:10.1214/ss/1009213286.