blocks|key|226447|text|zoo包中的rollapply接受任意函数。它与filter的不同之处在于，它在默认情况下排除了NA。|type|unstyled|depth|inlineStyleRanges|offset|length|style|CODE|entityRanges|data|226448|话虽如此，但为一个如此简单的函数加载一个包并没有多大意义(双关语)。|226449|这里有一个右对齐的：|226450|my.rollapply+<-+function(vec,+width,+FUN)+
++++sapply(seq_along(vec),+
+++++++++++function(i)+if+(i+<+width)+NA+else+FUN(vec[i:(i-width%2B1)]))

set.seed(1)
vec+<-+sample(1:50,+50)
my.rollapply(vec,+3,+sd)
+[1]++++++++NA++++++++NA++7.094599+12.124356+16.522712+18.502252+18.193405++7.234178++8.144528
[10]+14.468356+12.489996++3.055050+20.808652+19.467922+18.009257+18.248288+15.695010++7.505553
[19]+10.066446+11.846237+17.156146++6.557439++5.291503+23.629078+22.590558+21.197484+22.810816
[28]+24.433583+19.502137+16.165808+11.503623+12.288206++9.539392+13.051181+13.527749+19.974984
[37]+19.756855+17.616280+19.347696+18.248288+15.176737++6.082763+10.000000+10.016653++4.509250
[46]++2.645751++1.527525++5.291503+10.598742++6.557439

#+rollapply+output+for+comparison
rollapply(vec,+width=3,+sd,+fill=NA,+align='right')
+[1]++++++++NA++++++++NA++7.094599+12.124356+16.522712+18.502252+18.193405++7.234178++8.144528
[10]+14.468356+12.489996++3.055050+20.808652+19.467922+18.009257+18.248288+15.695010++7.505553
[19]+10.066446+11.846237+17.156146++6.557439++5.291503+23.629078+22.590558+21.197484+22.810816
[28]+24.433583+19.502137+16.165808+11.503623+12.288206++9.539392+13.051181+13.527749+19.974984
[37]+19.756855+17.616280+19.347696+18.248288+15.176737++6.082763+10.000000+10.016653++4.509250
[46]++2.645751++1.527525++5.291503+10.598742++6.557439|code-block|syntax|javascript|226451|entityMap^0|0|3|6|9|O|6|1C|2|0|0|0|0^^$0|@$1|2|3|4|5|6|7|Q|8|@$9|R|A|S|B|C]|$9|T|A|U|B|C]|$9|V|A|W|B|C]|$9|X|A|Y|B|C]]|D|@]|E|$]]|$1|F|3|G|5|6|7|Z|8|@]|D|@]|E|$]]|$1|H|3|I|5|6|7|10|8|@]|D|@]|E|$]]|$1|J|3|K|5|L|7|11|8|@]|D|@]|E|$M|N]]|$1|O|3|-4|5|6|7|12|8|@]|D|@]|E|$]]]|P|$]]

<code>rollapply</code> in the <code>zoo</code> package takes an arbitrary function. It's different from <code>filter</code> in that it excludes the <code>NA</code>s by default.

That being said, though, there's not much sense in loading a package for a function that's so simple to roll yourself (pun intended).

Here's one that's right aligned:

<pre><code>my.rollapply &lt;- function(vec, width, FUN) 
 sapply(seq_along(vec), 
 function(i) if (i &lt; width) NA else FUN(vec[i:(i-width+1)]))

set.seed(1)
vec &lt;- sample(1:50, 50)
my.rollapply(vec, 3, sd)
 [1] NA NA 7.094599 12.124356 16.522712 18.502252 18.193405 7.234178 8.144528
[10] 14.468356 12.489996 3.055050 20.808652 19.467922 18.009257 18.248288 15.695010 7.505553
[19] 10.066446 11.846237 17.156146 6.557439 5.291503 23.629078 22.590558 21.197484 22.810816
[28] 24.433583 19.502137 16.165808 11.503623 12.288206 9.539392 13.051181 13.527749 19.974984
[37] 19.756855 17.616280 19.347696 18.248288 15.176737 6.082763 10.000000 10.016653 4.509250
[46] 2.645751 1.527525 5.291503 10.598742 6.557439

# rollapply output for comparison
rollapply(vec, width=3, sd, fill=NA, align='right')
 [1] NA NA 7.094599 12.124356 16.522712 18.502252 18.193405 7.234178 8.144528
[10] 14.468356 12.489996 3.055050 20.808652 19.467922 18.009257 18.248288 15.695010 7.505553
[19] 10.066446 11.846237 17.156146 6.557439 5.291503 23.629078 22.590558 21.197484 22.810816
[28] 24.433583 19.502137 16.165808 11.503623 12.288206 9.539392 13.051181 13.527749 19.974984
[37] 19.756855 17.616280 19.347696 18.248288 15.176737 6.082763 10.000000 10.016653 4.509250
[46] 2.645751 1.527525 5.291503 10.598742 6.557439
</code></pre>

blocks|key|226491|text|TTR包中包含runSD：|type|unstyled|depth|inlineStyleRanges|offset|length|style|CODE|entityRanges|data|226492|>+library(TTR)
>+ls("package:TTR",+pattern="run*")
+[1]+"runCor"++++"runCov"++++"runMAD"++++"runMax"++++"runMean"++
+[6]+"runMedian"+"runMin"++++"runSD"+++++"runSum"++++"runVar"|code-block|syntax|javascript|226493|runSD将比rollapply快得多，因为它避免了进行许多R函数调用。例如：|226494|rzoo+<-+function(x,n)+rollapplyr(x,+n,+sd,+fill=NA)
rttr+<-+function(x,n)+runSD(x,+n)
library(rbenchmark)
set.seed(21)
x+<-+rnorm(1e4)
all.equal(rzoo(x,250),+rttr(x,250))
#+[1]+TRUE
benchmark(rzoo(x,250),+rttr(x,250))[,1:6]
#+++++++++++test+replications+elapsed+relative+user.self+sys.self
#+2+rttr(x,+250)++++++++++100++++0.58++++1.000++++++0.58+++++0.00
#+1+rzoo(x,+250)++++++++++100+++54.53+++94.017+++++53.85+++++0.06|226495|entityMap^0|7|5|0|0|0|5|7|9|0|0^^$0|@$1|2|3|4|5|6|7|Q|8|@$9|R|A|S|B|C]]|D|@]|E|$]]|$1|F|3|G|5|H|7|T|8|@]|D|@]|E|$I|J]]|$1|K|3|L|5|6|7|U|8|@$9|V|A|W|B|C]|$9|X|A|Y|B|C]]|D|@]|E|$]]|$1|M|3|N|5|H|7|Z|8|@]|D|@]|E|$I|J]]|$1|O|3|-4|5|6|7|10|8|@]|D|@]|E|$]]]|P|$]]

The TTR package has <code>runSD</code> among others:

<pre><code>&gt; library(TTR)
&gt; ls("package:TTR", pattern="run*")
 [1] "runCor" "runCov" "runMAD" "runMax" "runMean" 
 [6] "runMedian" "runMin" "runSD" "runSum" "runVar"
</code></pre>

<code>runSD</code> will be much faster than <code>rollapply</code> because it avoids making many R function calls. For example:

<pre><code>rzoo &lt;- function(x,n) rollapplyr(x, n, sd, fill=NA)
rttr &lt;- function(x,n) runSD(x, n)
library(rbenchmark)
set.seed(21)
x &lt;- rnorm(1e4)
all.equal(rzoo(x,250), rttr(x,250))
# [1] TRUE
benchmark(rzoo(x,250), rttr(x,250))[,1:6]
# test replications elapsed relative user.self sys.self
# 2 rttr(x, 250) 100 0.58 1.000 0.58 0.00
# 1 rzoo(x, 250) 100 54.53 94.017 53.85 0.06
</code></pre>

I know that the <code>filter()</code> function in R calculate the moving average. I would like to know if exists a function that return me the moving variance or standard deviation, in order to show it in a plot side by side with the output of <code>filter()</code> function.

Moving variance in R

我知道R中的filter()函数计算移动平均值。我想知道是否存在一个函数返回给我移动方差或标准差，以便在绘图中与filter()函数的输出并排显示。

问R中的移动方差
EN

社区

活动

资源

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问R中的移动方差EN

社区

活动

资源

关于

腾讯云开发者

热门产品

热门推荐

更多推荐

问R中的移动方差
EN