MAKE A MEME
View Large Image
| View Original: | Rule of three.svg (316x257) | |||
| Download: | Original | Medium | Small | Thumb |
| Courtesy of: | commons.wikimedia.org | More Like This | ||
| Keywords: Rule of three.svg en Image comparing the rule of three to the exact binomial one-sided confidence interval i e Clopper-Pearson 2014-07-10 09 39 11 own <br /> Created with Matplotlib Bscan <source lang 'python'> Python code shown below This code is issued under the Creative Commons CC0 License from __future__ import division import numpy as np import matplotlib pyplot as plt Set domain of the graph MaxTrials 10000 MinTrials 1 UCB np zeros MaxTrials Get points that are evenly spaced in log-space trials np exp np linspace np log MinTrials np log MaxTrials 100 One-sided exact binomial upper confidence bound equivalent to Clopper-Pearson The choice of confidence interval does matter but this is reasonable and is typically the one from which Rule of 3 is derived See Agresti and Coull 1998 for alternative binomial confidence intervals UCB 1-0 05 1/i for i in trials Rule_of_3 3/i for i in trials plt figure figsize 4 3 2 plt axes 0 17 0 13 0 79 0 8 plt hold True A a plt plot trials UCB '-' color 'orange' lw 2 5 b plt plot trials Rule_of_3 '-' color 'purple' lw 2 5 Formatting A append a A append b ax plt gca ax set_yscale 'log' ax set_xscale 'log' ax grid plt xlabel Sample Size plt ylabel r 95 Upper Confidence Bound leg_str leg_str append Exact Binomial leg_str append Rule of Three bx plt legend A leg_str numpoints 1 handletextpad 0 loc upper right plt savefig rule_of_three svg </source> cc-zero Uploaded with UploadWizard Statistics Probability | ||||
