I wish to perform moving tiles bootstrap resampling on some gridded data meteorological data. I've many years experience with S-Plus, but it has no way to perform a moving-tiles bootstrap. Within R I've learned how to use quadratresample() with the spatstats package and would be happy to simply use empirical percentiles if generating the replicates were fast, but it isn't. So, I'd like to employ bootstrap tilting to generate my confidence intervals. On my machine (older Athlon, XP sp3, 4 GB) it took about 48 h to generate 5000 moving tiles replicates. But, the boot package contains boot.tilt and I should be able to do well enough with the statistics I need (mean and RMS) to get by with several hundred samples instead of several thousand. I can certainly build a function that generates the statistic I want from a moving-tiles replicate. My problem is that the resampling process isn't as simple as boot.tilt() expects (I'm not simply resampling rows of a data frame). But, the function that does what I need doesn't fit into how boot.tilt() expects the resampling process to work. I can generate a string of replicates easily enough -- how might I wrap such a set of replicates into an object on which tilting could be performed absent the internal generation of the replicates themselves? Kim Elmore -- Kim Elmore, Ph.D. (CCM, PP SEL/MEL/Glider, N5OP, 2nd Class Radiotelegraph, GROL) /"Human beings, who are almost unique in having the ability to learn from the experience/ /of//others, are also remarkable for their apparent disinclination to do so/."/-- Douglas Adams/ [[alternative HTML version deleted]]
I wish to perform moving tiles bootstrap resampling on some gridded data meteorological data. I've many years experience with S-Plus, but it has no way to perform a moving-tiles bootstrap. Within R I've learned how to use quadratresample() with the spatstats package and would be happy to simply use empirical percentiles if generating the replicates were fast, but it isn't. So, I'd like to employ bootstrap tilting to generate my confidence intervals. On my machine (older Athlon, XP sp3, 4 GB) it took about 48 h to generate 5000 moving tiles replicates. But, the boot package contains boot.tilt and I should be able to do well enough with the statistics I need (mean and RMS) to get by with several hundred samples instead of several thousand. I can certainly build a function that generates the statistic I want from a moving-tiles replicate. My problem is that the resampling process isn't as simple as boot.tilt() expects (I'm not simply resampling rows of a data frame). But, the function that does what I need doesn't fit into how boot.tilt() expects the resampling process to work. I can generate a string of replicates easily enough -- how might I wrap such a set of replicates into an object on which tilting could be performed absent the internal generation of the replicates themselves? Kim Elmore -- Kim Elmore, Ph.D. (CCM, PP SEL/MEL/Glider, N5OP, 2nd Class Radiotelegraph, GROL) /"Human beings, who are almost unique in having the ability to learn from the experience/ /of//others, are also remarkable for their apparent disinclination to do so/."/-- Douglas Adams/ [[alternative HTML version deleted]]