When you run a extract database data via either a Portfolio Analysis, a Data Report or Portfolio | Project Data | Extract, you can control how that data is extracted with the Database Data Extraction Options dialog.
Create data files
When DPL queries the database connected to the portfolio, it can create DPL program files containing values and arrays with the data from the database. The data files can subsequently be used to run a Portfolio Analysis or Data Report. If you think you will be doing multiple runs or creating multiple Data Reports and are confident the data in the database will not change while doing your analysis, you may wish to check this box. Note: when you select Portfolio | Project Data | Extract, you are creating DPL programs so this check box is checked and disabled.
Data Sets
If the database is configured for revision tracking and/or with a review index, then DPL allows you to conduct analyses on specific revisions and/or review indices. When no boxes are checked within the section, DPL extracts the most recent revision for each data item.
If you'd like to extract an earlier revision, check the Extract earlier revision box and specify the date and time that the revision must come before. If, for example, the date is set to 3/1/2016 and time to 8:00:00 AM then DPL will extract the last revision of each data item that occurs at or before March 1st, 2016 at 8:00:00 AM. If for a given Node ID no such revision exists, then DPL will supply zeroes as missing data for that item (assuming you do not have Not allowed radio button selected in the Missing data section; if you do have Not allowed selected, then DPL stops extracting).
To extract a specific review index, check the Select Review Index bo and select the Review Period and Review Tag from the drop-down boxes. Depending on your selections within this section, you may have DPL extract the latest revision of a specified review index or previous revision of a specified review index. When the Select Review Index box is checked and the Extract earlier revision box is not, DPL will extract the latest revision of the review index specified. With both boxes in the Data Sets checked, DPL will extract the last revision before the specified revision date and time for the specific review index combination.
Missing data
This section of the dialog controls how DPL will behave if data that it expects to find is missing or inconsistent. The three settings and their meanings are as follows:
- Node data override. This setting solely uses the information in the node definition to define the data.
The setting ignores inconsistencies between node definition dimension information and database dimension information. It also automatically supplies missing values (a zero) as needed. For example, assume there is value node called Value1 in the influence diagram that is defined to be a 3-row x 4-column array. If this were initialized with a database link that was a record that contained data for a 1-row x 4-column array, DPL would create a data structure that was a 3 x 4 array with the data from the database for the first row and zeroes for the remaining three rows.
DPL does not issue an error message if there is an inconsistency between the node definition and the database information or if there is missing data. If there is an inconsistency or missing data, a message is written to the Session Log.
This setting is useful if the data in the database is not completely specified.
- Inconsistencies not allowed. This setting forces the node definition and database definition dimension information to match.
DPL prompts you once as to whether you would like it to supply missing values (zeroes) if data is missing. If you respond to allow missing data, DPL will not prompt again and will continue supplying zeroes as needed for missing data. However, if an inconsistency between dimension information exists such as in the example above, DPL will not continue extracting data.
For example, assume there is value node called Value2 in the influence diagram that is defined to be a 3-element array (i.e, 1-dimensional). If this were initialized with a database link that was a record that contained data for 3-element array but that only supplied data for two elements, DPL would create a data structure that was a 3-element array with the data from the database for the first two elements and a zero for the last.
DPL does not issue an error message if there is missing data. It prompts once and then continues if you tell it to. If there is missing data, a message is written to the Session Log. If there is an inconsistency between the node definition and the database dimension information, DPL gives an error message and stops extracting.
- Not allowed. This is the most restrictive setting. The database data must be fully specified and it must match node data definition.
If DPL encounters missing data or an inconsistency in dimension information, DPL gives an error message and does not continue extracting data.
Versions: DPL Portfolio
See Also
Select Input dialog (Data Report)