## The mixed frequence problem of Winrats

Use this for suggestions about improvements in RATS
hardmann
Posts: 193
Joined: Sat Feb 26, 2011 9:49 pm

### The mixed frequence problem of Winrats

Dear Tom:

In Winrats, the calendar() instruction to determine the frequency and starting point of the time series, allocate() instruction determines the endpoint of a time series. These previous config the built-in timeline or Time reference ruler, then facilitate the analysis of time series. In Matlab and Python, there is no corresponding Time reference ruler, and time series analysis is conducted using matrices. A separate matrix, similar to a Time reference ruler, is established to specifically identify time. Therefore, the mixed frequence is carried out using different matrices and their respective time ruler. In Eviews, it uses a timeline method similar to Winrats, but it uses different pages to set different frequencies, and achieves mixed frequence operations through frequency correlation on the pages.
When using mixed frequence in Winrats, low-frequency data is usually placed in its corresponding position in high-frequency data, such as in quarters/months, placing quarterly values in months 3, 6, 9, and 12. Place the annual value in the fourth quarter of the year/quarter. However, in quarterly/monthly setting, the analysis of quarterly data cannot be carried out and can only be done in other rpf file.
It is recommended to implement dual or multiple Time reference rulers in a rpf programming file in Winrats, similar to multithread or subspace in computer software. Different frequency time series reference different Time reference ruler. This way, high-frequency and low-frequency data can be analyzed simultaneously, and mixed frequence calculation can be directly performed through the association of Time reference ruler.

Best regard
Hardmann
TomDoan
Posts: 7627
Joined: Wed Nov 01, 2006 4:36 pm

### Re: The mixed frequence problem of Winrats

I'm not sure what it is you think you can't do. You can switch between CALENDAR schemes at will:

Code: Select all

``````open data demodisaggregate.rat
calendar(q) 1929:1
data(format=rats,compact=average) 1929:01 2012:04 indpro gdpc1
compute asQuarterly=%calendar()
calendar(a) 1929:1
data(format=rats) 1929:1 2012:1 gdpca
compute asAnnual=%calendar()
*
calendar(recall=asQuarterly)
graph
# indpro
calendar(recall=asAnnual)
graph
# gdpca
``````
Could you be (much more) specific about what you would like to do?