Non Gregorian Calendar

[abi]

Dear Tom,

Does RATS match with Non Gregorian Calendar? I tried it with the annual and seasonal data and there was no problem, but it does not seem to work properly with the daily data.

[TomDoan]

There are many different non-Gregorian calendars. The RATS date functions all assume a Gregorian calendar dating back to 1 C.E. which isn't correct for very old daily data (depending upon when a country switched from Julian to Gregorian). If you're talking about a lunar calendar, there are many of those and most are quite complicated.

[abi]

Thanks for reply Tom,

I would like to use Persian calendar. I 'm using following code to read data but it seems does'nt work properly. The data (attached) started

Code: Select all

OPEN DATA "E:\ARCHIVE-INDEX\DATA3.XLSX"
CALENDAR(D) 1394:9:2
DATA(FORMAT=XLSX,ORG=COLUMNS,dateformat="yyyymmdd") 1394:09:02 1399:09:02 KHD MAINMTL

but RATS considers data from 1394:09:02 to 1399:04:03. Is there any way to fix this problem? It should be noted in Persian calendar thursday and friday are weekly official holidays and also there are some occasional holidays throughout the year.

[TomDoan]

What is the problem that you want to "fix"? You have (very) irregular daily data. What part of your analysis would depend upon knowing the date of a particular entry?

[abi]

Well, at present the correct date is not shown on the horizontal axis of some graphs.

[TomDoan]

The version 10 addition of the irregular calendars would help somewhat. It would still remap some of those dates---I noticed that the 2 month sometimes has 31 days, which would be translated into a March 2 or 3 so two dates might be converted to the same Julian value. However, your data has so many fairly long gaps that you should just live with sequence numbers and perhaps use grid lines to show year breaks.

[abi]

The version 10 addition of the irregular calendars would help somewhat.

It mean's if i update to version 10 then the problem would be fix?

[TomDoan]

It would give you something closer to what you might want, but it still has no idea how to handle "February" with 31 days---it will make a few errors like that. If you have enough time periods that it's only doing to be able to label years anyway, then you wouldn't notice any one or two day misses.