[R] cumulative growth rates indexed to a common starting point over n series of observations

[Dirk Eddelbuettel]

On 1 September 2006 at 14:30, toby_marks at americancentury.com wrote:
| The apply with the cumprod was exactly what I was after.  The apply just 
| wasn't clicking with me, and I had overlooked the cumprod.  Thanks to all 
| for pushing me down the right path!
| 
| Actually, what I am ultimately after is a way to link this series, without 
| having to use a for loop ( the only way I can think of ... ).  But, would 
| like to see if it can be linked  using mapply or apply against the rows 
| and to compute the linked results. 
| 
| zz = matrix(rnorm(20), ncol=2)
| zzcum = apply(zz/100 + 1, 2, cumprod)
| zzlinkcum = 100*zzcum
| for(i in 2:length(zz[,1])){ zzlinkcum[i,]=zzlinkcum[i-1,]*zzcum[i,]}  ### 
| Is there a better way here ?

Sure, why not call apply again?

> set.seed(42); zz <- matrix(rnorm(20), ncol=2)
> zzcum <- apply(1+zz/100, 2, cumprod)
> apply(rbind(c(1,1), zzcum), 2, cumprod)*100
          [,1]     [,2]
 [1,] 100.0000 100.0000
 [2,] 101.3710 101.3049
 [3,] 102.1804 104.9735
 [4,] 103.3704 107.2642
 [5,] 105.2360 109.2994
 [6,] 107.5684 111.2246
 [7,] 109.8358 113.9036
 [8,] 113.8461 116.3156
 [9,] 117.8912 115.6233
[10,] 124.5442 112.1301
[11,] 131.4900 110.1781
> 

Hth, Dirk

-- 
Hell, there are no rules here - we're trying to accomplish something. 
                                                  -- Thomas A. Edison