Re: Re: st: Filtering methods with short time series

[Nick Cox <njcoxstata@gmail.com>]

Independently of this kind of issue, almost all these filters are just
some kind of weighted average. Weighted averages sometimes disappoint
their users if the data are strongly skewed, so some kind of
transformation first may help a bit.

Nick

2011/2/1 Jorge Eduardo Pérez Pérez <perez.jorge@ur.edu.co>:

> After installing the mFilter package in R, if one writes down:
>
> require(mFilter)
> cffilter
>
> the source code for the routine is displayed.


> On Mon, Jan 31, 2011 at 7:49 PM, Christopher Baum <kit.baum@bc.edu> wrote:
>> <>
>> Jorge said
>>
>> Try to avoid -cfitzrw-, I have compared the output from
>> implementations of the Christiano Fritzgerald filter in other software
>> to the Stata output and they are different. I already contacted the
>> authors about this, but I haven't received any response.
>>
>>
>> Jorge raised this issue with me a couple of weeks ago. The Stata -cfitzrw- routine is, I believe, a faithful translation of a MATLAB
>> routine on the web. He also contacted the author of the MATLAB routine, but I guess he has not responded to the enquiry.
>>
>> Jorge pointed to differences between the Stata results and results obtained from a R implementation and a Excel implementation.
>> The R implementation, as far as I can see, does not contain readable source code. I can run the routine, but I cannot analyze its
>> workings to determine what it may be doing differently than -cfitzrw-. The Excel routine has code that I can read, but cannot execute.
>> I will try to figure out what the differences may be based on that latter routine, but without the ability to run it (or the ability to view
>> the source code of the R routine) it is rather difficult to debug what flaws might exist in the logic of -cfitzrw-.

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/