[SciPy-User] how to use kendalltau_seasonal

[Pierre GM]

On Apr 6, 2011, at 12:41 PM, Andreas wrote:

> Hi there,
> I'm a bit at a loss how to use scipy.stats.mstats.kendalltau_seasonal.
> The docstring and the online documentation aren't really that helpful
> about what input and output mean ...

<sigh>Somebody was too busy coding to pay attention to doc writing...</sigh></sorry>

Imagine you need to check whether there's a trend in a time series X, but want to take a potential seasonal dependence into account.
Reshape your data so that it's 2D, with seasons as columns, and feed it to the function. You'll get a dictionary giving you the value of Kendall's tau (w/ and w/o the seasonal effects) with the corresponding p-values that should help you figure out without there's a trend in your data and whether seasonality plays a role.
Example:
You have a 60-element array corresponding to 5 years of consecutive monthly data, and want to check the trend, taking potential monthly variations into account. In that case, your "season" is a month. Reshape it as a (5,12), and use the result as input to the function.

More info:
http://pubs.usgs.gov/sir/2005/5275/pdf/sir2005-5275.pdf
Hirsch R.M., Slack J.R. and Smith R.A. (1982), Techniques of Trend Analysis
for Monthly Water Quality Data, Water Resources Research 18(1), 107-12

I'm sure there must be a R equivalent somewhere.
In any case, please do help to document the function as needed. Don't hesitate to contact me if you need more info.
Cheers
P.