[R-SIG-Finance] Ranks of Spearman's rank correlation coefficient

[Gabor Grothendieck]

On Sun, Jan 9, 2011 at 8:22 AM, Zanella Marco <marco.zanella at inbox.com> wrote:
> Dear Sirs,
> I'm using cor() funcion with Spearman method to analyse correlation of timeseries. As you know Spearman method is based on ranks of timeseries, instead of their value.
>
> Well, what I'm not able to do is to extract the ranks of the timeseries used from cor() to compute the Spearman correlation coefficient (also called rho).
>
> For instance if I have this two timeseries
>> a = c(-0.10, -0.21, -0.35, -0.38, -0.19)
>> b = c(-0.53, -0.42, -0.59, -0.55, -0.35)
>
> I obtain rho in this way
>> cor(a,b, method="spearman")
>>[1] 0.6
>
> but how can I obtain ranks for the two timeseries? Of course, the same computed and used by previous cor().
>

Try this:

> cor(rank(a), rank(b))
[1] 0.6


-- 
Statistics & Software Consulting
GKX Group, GKX Associates Inc.
tel: 1-877-GKX-GROUP
email: ggrothendieck at gmail.com