|
|
| |
VEDA (VErsatile Data Analyst)
A versatile tool is evolving to satisfy the needs of numerous modelers and analysts confronted with complex models requiring extensive manipulation of input data and results. VEDA is a powerful yet user friendly set of tools geared to facilitate the creation, maintenance, browsing, and modification of the large data bases required by complex mathematical and economic models, as well as the smart exploration of the results created by such models and the creation of reports. The VEDA system is composed of two major subsystems:
Results handling:
The VEDA results Analyzer (VEDA back-end or VEDA-BE): VEDA-BE is a powerful result analyser that constructs tables and graphs from the results obtained from complex mathematical and economic models. Version 3.7 is customized for MARKAl and TIMES models, whereas Version 4.0 is fully customizable by the user to his/her own model.
A common characterisitic of all complex models is that its results involve a large number of dimensions. Examples of dimensions that are common to most models are: time, scenario, process, and flows. However, each model has its own set of dimensions which may be quite large. MARKAL typically has 8 dimensions, while some complex transshipment models may have many more. As an example, the EUGENE model for Solid Waste Management has 24 dimensions.
While a single result table must eventually be printed in the two dimensions of a paper sheet, it may still involve more than two dimensions, via nested rows and/or columns. The user is confronted to a very large number of possible types of tables, both concerning the contents of the table, and its physical layout. VEDA-BE makes these choices simple and time efficient via a technology that combines powerful databases, intuitive screens, and the extensive use of drag-and-drop tools.
VEDA-BE is used by the MARKAL community for extensive reports, and its new version is being disseminated to other models' users.
Input handling:
As one can imagine, it is hard to do without domain specific knowledge on the input side, so we don't have a ready-to-use tool here. However, we specialize in rapid building of customized solutions, which usually includes a set of spreadsheets, backed up by a database system. Our stated goal is "to take the drudgery out of modeling", and we are guided by the following principles:
· Distinguish between data and assumptions
o Data
§ We build direct conduits to the source, as far as possible. The most important advantage of doing this is incorporating periodic updates of the source data.
o Assumptions
§ They should be expressed in an intuitive form
§ They should be easy to change
· Ability to view the data in a database environment, with the same filtering and configuring capability as the results analyzer.
|
|
|
