In a previous post we discussed about the possibility of using Machine Learning to build up a novel complexity index (called MONEY, see here: https://www.nature.com/articles/s41598-022-13206-0). The MONEY index is based on the prediction of the Relative Comparative Advantage (RCA) of countries based on a machine learning tool called Matrix Completion. In this work, we start from the MONEY index to understand how much each country in the dataset "helped" in the prediction of other countries' RCA (and hence in the construction of the index).
This exercise is useful to understand the relationship between countries and to check if traditional measures of similarity (e.g. cosine similarity) really capture the affinities of countries in terms of relative advantages and disadvantages in the export of goods.
If you are curious about how Matrix Completion can help unveiling the similarities in export experience of countries, please check out our published paper here.