phcsoares / econci Goto Github PK

Hi,

Thank you for your effort to implement the ECI algorithm!

However, I do have a question that needs your help. After calculating the ECI for each country and the PCI for each product, why there is an additional step to "fix the sign"? I know the rankings may be reversed without this step, but I can't find it in the papers or books by the authors. Is it a typical procedure in linear algebra?

In addition, according to Kemp-Benedict (2014), ECI and the diversity of countries are orthogonal. Does it mean that ECI is independent of diversity and should have a close-to-zero correlation coefficient? In my experiment with your and the official code, ECI and the diversity show a significant positive correlation (~0.5, both Pearson's and Spearman's correlation). Do I misunderstand what orthogonal means?

Thank you!

The current implementation can not handle very large matrices. One solution is to perform all computations in scipy sparse matrices.

Hello, the method you provided is of great help to me. Thank you for your contribution. I've had some problems with computational data builds while working with your code. If it is convenient for you, can you provide an example of data? Thank you very much again. Best Wish !

Hi Pedro, I could only find the RCA score in the list. Is it possible to do an algorithm for the Fitness score as well?
Check this for how to calculate:

https://journals.plos.org/plosone/article/file?id=10.1371/journal.pone.0117174&type=printable