Hi Guys,
I have been working with a weekly panel dataset and created several dbs to hold the data. One thing I discovered is that when you store data in database, instead of showing "A" or "W" or "Q" for frequency, the panel variables have a "U" for undated (?). Should it be something like a "P" for panel and then "W" for week to indicate what it actually is?
Tchaithonov
DB Frequency
Moderators: EViews Gareth, EViews Moderator
-
EViews Chris
- EViews Developer
- Posts: 161
- Joined: Wed Sep 17, 2008 10:39 am
Re: DB Frequency
I'm afraid that EViews databases don't really understand panel data series.
When you store a series from a panel into a database, the panel structure is discarded and the series is treated as a simple undated series.
If you import the series back into exactly the same panel workfile, things will work fine, but fetching it into any other workfile will probably not produce the results you might expect.
What's going on is that a series in an EViews database contains only the data values of the series and some simple meta information on frequency and start date. To store a panel series properly, we would also need to store the complete set of cross section identiers and time series identifiers for every observation. We haven't done things this way partially due to a lack of demand from users for working with panel series in databases, and partially because it will make for a fairly large database (since if you store 100 panel series you will end up with 100 full copies of the panel identifier series).
I think that explains the 'U', although it might not be the answer you hoped for.
When you store a series from a panel into a database, the panel structure is discarded and the series is treated as a simple undated series.
If you import the series back into exactly the same panel workfile, things will work fine, but fetching it into any other workfile will probably not produce the results you might expect.
What's going on is that a series in an EViews database contains only the data values of the series and some simple meta information on frequency and start date. To store a panel series properly, we would also need to store the complete set of cross section identiers and time series identifiers for every observation. We haven't done things this way partially due to a lack of demand from users for working with panel series in databases, and partially because it will make for a fairly large database (since if you store 100 panel series you will end up with 100 full copies of the panel identifier series).
I think that explains the 'U', although it might not be the answer you hoped for.
-
tchaithonov
- Posts: 168
- Joined: Mon Apr 13, 2009 7:39 am
- Location: New York City
Re: DB Frequency
So I guess the only way is to pre-treat the data before putting them into panel, similar to what I put down in this post
http://forums.eviews.com/viewtopic.php? ... &sk=t&sd=a
Well then.. thanks anyway Chris.
http://forums.eviews.com/viewtopic.php? ... &sk=t&sd=a
Well then.. thanks anyway Chris.
Return to “Suggestions and Requests”
Who is online
Users browsing this forum: No registered users and 1 guest
