Counting dates between observations

[mikela]

Hi everyone,
I have a simple but tedious problem to solve and i am starting to feel desperate...

I have lots of times series (annual data from 1900-2011) with lots of #NAs in them. What i need to do is to produce a table with the number of years between observations.
Is there an easy way to do this in eviews? I was thinking about @datediff but, for a start, I don't know how to pick the date value (for the existing observations)!

Any help is very much appreciated,
thanks
Mikela

[EViews Gareth]

Something like this:

Code: Select all

smpl @first @first series count=x=na smpl @first+1 @last count = @recode(x=na, count(-1)+1,0)

[mikela]

thanks! i was struggling with cumsum (because it continue to cumulate regardless of the missing values) but with this code was much easier, thanks again!

[Jen]

Hi,

I have time series data for 2001, 2003, 2005, and 2007, which workfile structure type do I select?

I have selected unstructured/undated but it does not seem to be working.

What is the minimum number of observations needed for time series data?

Many Thanks,
Jen

[EViews Gareth]

Dated - regular frequency

Frequency - Multi-year (2 year)

Start date: 2001
End date: 2007

[Jen]

Thanks Gareth that really helps.

Do you have to have data at regular intervals, for example could i have 2001, 2003, 2005, 2006, 2007, 2008?

Also what is the minimum number of observations required to run White Heteroskedacity cross terms test?

Many Thanks
Jen

[EViews Gareth]

If you want irregularly spaced data, you'll have to create an irregular workfile. You'll have to consult the documentation for more information.


There is never a correct answer to the "how many observations do I need" question.