The accuracy score reported should be around $71\%(0\mathrm{.}71).$ This means that approximately $29\%$ of the countries in the test set had their regions mispredicted. While that doesn't sound great, it could be that the WHR numerical data are not always completely predictive of region. One could imagine some countries that are "outliers" in a particular region, and more closely resemble other regions based on the WHR indicators.

In the below code cell, please loop over all the predicted and true values in the test set, and prints out the country name and predicted region when the prediction is incorrect. An output line like: Sri Lanka : South Asia $->$ Sub$-$Saharan Africa means that Sri Lanka is actually part of the South Asia region but was predicted to be part of Sub$-$Saharan Africa.

python code