[R] Internal NA removal out of Time Series with na.omit.ts()

[Adrian Trapletti]

Prof Brian Ripley wrote:

>On Mon, 8 Mar 2004, Adrian Trapletti wrote:
>
>  
>
>>>The na.omit.ts() method fails when the time series contains internal
>>>NA's. How can these automatically be removed?
>>>      
>>>
>>try na.remove from tseries. This is, e.g., useful when removing weekends 
>>(NA prices) from financial data, i.e., switching from physical time to 
>>business time.
>>    
>>
>
>That may well not be appropriate though.  NA denotes a missing and not a 
>non-existent value, 
>
I think, this is more a philosophical question, if weekends represent 
missing or non-existent values. One can find arguments for both.

>and for example removing non-trading days in just one 
>market it probably not appropriate (as information accrues in other 
>markets).
>  
>
>The issue is not `How can the na.action be activated correctly?', but
>`What is the appropriate na.action?', and the answer is usually `none of 
>them'.
>
>  
>
I partly agree with you. Sometimes it is not appropriate and it might be 
better to use, e.g., a state-space model which can work with NA's. But 
from my experience in financial markets, it is often not worth 
considering these more complex approaches, since they don't generate 
additional profits. Other effects often dominate...
But anyhow, I find it good if the user can choose between the different 
options and decide himself.

>>>>>spectrum(ts.mNDII, na.action=na.omit)
>>>>>          
>>>>>
>>>Error in na.omit.ts(as.ts(x)) : time series contains internal NAs
>>>
>>>How can the na.action be activated correctly?
>>>      
>>>
>
>  
>
best
Adrian

Dr. Adrian Trapletti
Trapletti Statistical Computing
Wildsbergstrasse 31, 8610 Uster
Switzerland
Phone & Fax : +41 (0) 1 994 5631
Mobile : +41 (0) 76 370 5631
Email : mailto:a.trapletti at bluewin.ch
WWW : http://trapletti.homelinux.com