Usages of Bootstrap Confidence Intervals R Functions

  1. One-Sample Mean
         res <- bootInterval(x) # 95% bootstrap intervals based on 1000 bootstrap samples
         check(res) # check normality of bootstrap samples
         summary(res) # 95% bootstrap percentile confidence intervals
         # If normality holds then can do
         summary(res, "t") # 95% bootstrap t confidence intervals
         summary(res, "both") # both percentil and t intervals
         summary(res, "b") # same as the above
         res < bootInterval(x, conf.level=0.90)  # 90% confidence interval(s)
         res < bootInterval(x, B=2000)  # bootstrap size is 2000
        
  2. One-Sample Variance
         res <- bootInterval(x,par="variance") # for variance
         res <- bootInterval(x,par="v") # same as the above
         res <- bootInterval(x,par="variance",log=T) # for log(variance)
        
  3. One-Sample Proportion
         res <- bootInterval(x,par="proportion") # for proportion
         res <- bootInterval(x,par="p") # same as the above
        
  4. Difference in Two-Sample Means
         res <- bootInterval(x,y) # for mu1 - mu2
         # Note: bootstrap percentile confidence intervals only
        
  5. Ratio of Two-Sample Variances
         res <- bootInterval(x,y,par="v",comp="r") # for sigma1^2 / sigma2^2
         res <- bootInterval(x,y,par="v",comp="r",log=T) # for log(sigma1^2 / sigma2^2)
        
  6. Two-Sample Proportions (using summary statistics)
         # Summary statistics are
         # sample 1: 120 out of 200,  sample 2: 73 out of 100
         res <- bootInterval(120,73,200,100,par="p",comp="d") # for p1 - p2
         res <- bootInterval(120,73,200,100,par="p",comp="r") # for p1 / p2
         res <- bootInterval(120,73,200,100,par="p",comp="r",log=T) # for log(p1) - log(p2)
        
  7. Paired-Sample for Means
         res <- bootInterval(x,y,paired=T)
        

Feb. 21, 2008