用“增长"滚动应用窗户 [英] rollapply with "growing" window

问题描述

伙计们，通常当您执行以下操作时:

Guys, normally when you do something like:

tmp = zoo(rnorm(100), 1:100)
rollapply(tmp, 10, function(x) quantile(x, 0.05), align="right")

完全正确 rollapply 将从 10 个元素可用的那一刻开始计算值.

Quite rightly rollapply will start calculating the value from the moment 10 elements are available.

不幸的是，我需要在前 10 个观察中使用尽可能多的数据的东西，本质上是一个不断增长的数据窗口，直到有足够的数据来使用滑动窗口，例如1, 1:2, 1:3, 1:4 等，直到我们有至少 10 个元素，然后像往常一样滑动窗口.

Unfortunately I need something that uses as much data as possible for the fist 10 observations, essentially a growing window of data till there is enough data to use a sliding window, e.g. 1, 1:2, 1:3, 1:4, etc. till we have at least 10 elements and then slide the window as usual.

还有比丑陋的 for 循环更好的方法吗?

Is there a better way to do this than an ugly for loop?

推荐答案

为什么不一开始就用 9 个 NA 填充系列?绝对比丑陋的 for 循环"好:

Why not just pad the series with 9 NAs at the beginning? Definitely better than "ugly for loops":

tmp = zoo(c(rep(NA,9), rnorm(100)), 1:109)
zoo(rollapply(tmp, 10, function(x) quantile(x, 0.05, na.rm = TRUE), 
              align="right"), 1:100)

这篇关于用“增长"滚动应用窗户的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！