GTAP Resources: Resource Display
| GTAP Resource #3807 |
|---|
|
"MachiLearning as a data driven tool in result analysis" by Britz, Wolfgang Abstract Simulations with large-scale economic models generate huge amount of numerical data, based on a high differentiation by sectors/products, in space, by policy instruments. Lately, large-scale sensitivity analysis or using stochastic draws has come into fashion which let result sets explode in size. Besides the computational challenges related to managing large-scale data sets, the analyst faces the challenge to dig up a story out of that haystack. What results are worth reporting? How can they be explained by cause-effect relations? What regions / sectors / institutions are winners or losers and why? Since a long time, modelers have developed post model tools to e.g. aggregate results, derive indicators and decompose results. These approaches require a fair amount of priori knowledge and a priori decisions about what aspects to include in the analysis and how link them. Especially if new features are added to a modeling system, such as the new environmental satellites in GTAP, or in case of large-scale sensitivity analysis, the a priori knowledge might not be sufficient to define indicators and decompositions approaches which capture the main message. Here, more data driven approaches might help. A relatively new and so far not widely used and explored instrument for result analysis is the integration of a machine learning, such a software package has been recently integrated in the CAPRI GUI. The idea is to use data driven approaches which do not require (strong) ex ante knowledge about cause and effect relations underlying the results. The paper discusses that appproach. |
| Resource Details (Export Citation) | GTAP Keywords | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
- Software and modeling tools |
| Attachments |
|---|
|
If you have trouble accessing any of the attachments below due to disability, please contact the authors listed above.
Public Access  Machine Learning as a data driven tool in result analysis (1.1 MB) Replicated: 0 time(s)Restricted Access No documents have been attached. Special Instructions No instructions have been specified. |
| Comments (0 posted) |
|---|
|
You must log in before entering comments.
No comments have been posted. |
Last Modified: 9/15/2023 1:05:45 PM
