Category:Index: Difference between revisions
Jump to navigation
Jump to search
(first draft) |
(new ideas based on own thinking) |
||
| (One intermediate revision by the same user not shown) | |||
| Line 1: | Line 1: | ||
[[Category:Dimension]] | [[Category:Dimension]] | ||
{{index}} | |||
This page contains [[Index|indices]]. | This page contains [[Index|indices]]. | ||
==Scope== | |||
How should the indices be operationalised so that | |||
* when you load a model or data, you don't need to think whether the index exists already. | |||
* The indices that repeat the same theme (e.g., age) can be handled in a practical way without large management (e.g. no need for a separate wiki page for each individual index)? | |||
==Result== | |||
* All age indices (for example) may be put into one wiki page and use the same Ident. | |||
** These indices are treated as one index in Opasnet, although they are separate indices in Analytica. | |||
* The Ident is descriptive, such as Age, Sex, Year and so on. This can be used also as identifier in Analytica. | |||
** If the Ident of an index is omitted, the identifier is used. Note! All variables must have Idents like Op_en1912. (Why?) | |||
* The set memberships of indices in dimensions are described afterwards, so the data uploader does not need to worry about that. | |||
* The index wiki pages and the wiki page names to the Base are added afterwards by a moderator, so the data uploader does not neet to worry about that. | |||
* If an index exists but does not have the locations that the variable uses, the new locations are simply added. | |||
* Sets and items are not filled at all, because it is not crucial. | |||
Problem: if you can have any number of rows for every index, it becomes unclear how these rows are ordered. How do you do this? | |||
* If two rows are equal, do you want to use the same row number to those? If yes, how do you know which one is preferable? | |||
** There must be some preferred row among equal rows, like Male and Female are preferred over 1 and 2. | |||
*** The preference can vary depending e.g. on the language which with you operate? | |||
Temporary conclusions: The idea of non-fixed-row indices makes the system so much flexible in the upload and interpretation phase that I would go for that even if the download phase is unclear. | |||
* There is probably a need for moderator management. The moderator tells the system which rows are equal, so the user does not need to think about that. | |||
Latest revision as of 14:47, 26 March 2009
| Metadata for this index (please use these attributes in Analytica and Result database) | |
| Ind_name (identifier in Analytica) | Wik_id |
| Op_en2664 | 1 |
Category:Index]] This page contains indices.
Scope
How should the indices be operationalised so that
- when you load a model or data, you don't need to think whether the index exists already.
- The indices that repeat the same theme (e.g., age) can be handled in a practical way without large management (e.g. no need for a separate wiki page for each individual index)?
Result
- All age indices (for example) may be put into one wiki page and use the same Ident.
- These indices are treated as one index in Opasnet, although they are separate indices in Analytica.
- The Ident is descriptive, such as Age, Sex, Year and so on. This can be used also as identifier in Analytica.
- If the Ident of an index is omitted, the identifier is used. Note! All variables must have Idents like Op_en1912. (Why?)
- The set memberships of indices in dimensions are described afterwards, so the data uploader does not need to worry about that.
- The index wiki pages and the wiki page names to the Base are added afterwards by a moderator, so the data uploader does not neet to worry about that.
- If an index exists but does not have the locations that the variable uses, the new locations are simply added.
- Sets and items are not filled at all, because it is not crucial.
Problem: if you can have any number of rows for every index, it becomes unclear how these rows are ordered. How do you do this?
- If two rows are equal, do you want to use the same row number to those? If yes, how do you know which one is preferable?
- There must be some preferred row among equal rows, like Male and Female are preferred over 1 and 2.
- The preference can vary depending e.g. on the language which with you operate?
- There must be some preferred row among equal rows, like Male and Female are preferred over 1 and 2.
Temporary conclusions: The idea of non-fixed-row indices makes the system so much flexible in the upload and interpretation phase that I would go for that even if the download phase is unclear.
- There is probably a need for moderator management. The moderator tells the system which rows are equal, so the user does not need to think about that.