D4.5 Outlier management and reporting
Updated: Apr 20, 2020
This deliverable addresses the latest progressof the SMEthod decision support tool (DST). In particular, we will discuss here below the following topics:
1.the re-calibration functionality of the DST as new knowledge enters the system;
2. the support of restricted instances. By 'restricted instances' this we imply the important concept of sub-models, which are based on restricted views of the dataset and not on the whole data set. The importance of this is paramount as one can in this way, cater for regional (and other) particularities. Thus, one may end up with an EU model as well as with several regional models;
3. The issue of outlier management, i.e., how we can define outlier data among our dataset, that may be corrupting and degrading our models essentially; In addition, we will also present an important part of the DST. 4. the reporting functionality, whereby results are packaged in custom pdf reports;
In the end, we will provide for a critical reappraisal of the overall approach from the stakeholder point of view, addressing the perhaps the most important aspect of the whole project.
5. what is the way forward, including suggestions raised about introducing to the DST a function enabling particular stakeholders to assign weights to selected variables.
Download the full report (PDF) here.