Hi Florian (Tester Trying):

I would agree with Rohit on needed clarifications. The initial source data file and UN data in csv format need cleaning. The data files contain some incongruences pertaining to WolframData and the nature of country names over time and multiple regions with undefined sub-regions. As per your title, some pivoting on the M49 code helps. The following code is much the same as before up to UNcountryprop where we diverge away from Association and Dataset and just manipulate list. Instead of running through each line of the "master.csv" file and adding the two elements (columns) {"continent", "region"} it helps to optimize by reducing to a set of distinct countries, then mapping {"continent", "region"}

Clear[master, mastercountriesraw, UNmethoddata, addcountryprop, UNcountryprop, addcontinentandregionforexport, masterout, headings];
master = Import["C:\\suicide-rates-overview-1985-to-2016\\master.csv"];
mastercountriesraw = DeleteDuplicates[Rest[master[[All,1]]]];
UNmethoddata = Import["C:\\suicide-rates-overview-1985-to-2016\\UNSD — MethodologyFixed.csv"];
addcountryprop[cnty_] := Module[
       { countryraw, countryent, continent, unnumber, globalcode, globalname, regioncode, regionname, subregioncode, subregionname, intermediateregioncode, intermediateregionname, countryorarea, M49code, bdata}, 
       countryraw = cnty;
        countryent = Quiet[CountryData[cnty]];
        If[ToString[Head[countryent]] == "CountryData", countryent = Interpreter["Country"][cnty]];
        continent = CanonicalName[CountryData[countryent, "Continent"]];
        unnumber = CountryData[countryent, "UNNumber"];
        { globalcode, globalname, regioncode, regionname, subregioncode, subregionname, intermediateregioncode, intermediateregionname, countryorarea, M49code } = 
       Take[SelectFirst[UNmethoddata, (#[[10]] == unnumber) &], 10];
        bdata = Select[master, (#[[1]] == countryraw) &];
        { globalname, regionname, subregionname, If[intermediateregionname == "", subregionname, intermediateregionname], continent, countryraw, Rest /@ bdata }
       ];
UNcountryprop = Map[addcountryprop[#] &, mastercountriesraw];
addcontinentandregionforexport[item_] := Module[
       { contiregcountr }, 
       contiregcountr = {item[[5]], item[[4]], item[[6]]};
        Map[Join[contiregcountr, #] &, item[[7]] ] 
       ];
masterout = Flatten[Map[addcontinentandregionforexport[#] &, UNcountryprop], 1];
headings = Join[ {"continent","region"}, First[master] ];
Export["C:\\suicide-rates-overview-1985-to-2016\\masterout.csv", masterout, "CSV", "TableHeadings" -> headings]

In reviewing previous code I spotted some issues with the GroupBy; but you don't want a Dataset .Using the Wolfram Language to add two columns to a CSV file is actually straight forward. There are some complications as to how to add those columns in your post questions.

Hi Tester Trying,

Can you be more specific regarding

I tried both versions but it didn't work for me

What exactly did not work? Was an error reported?

My answer does exactly what you requested, add continent and region columns. The raw data was imported as a Dataset because it provides a lot of useful functionality for working with structured data. I would suggest learning how to use Dataset, take a look at the examples.

If you prefer working with CSV outside of Mathematica then you can always Export the dataset as CSV.

Dear Hans Michel,

I have this dataset with all countries and multiple entries for different years per country. Ideally I would like to add two kind of rows. One with the continents and one with geographical regions like 'Caribbean', 'Central America' or 'North/South Europe' for example.

dataset: https://www.kaggle.com/russellyates88/suicide-rates-overview-1985-to-2016

regions:

Hello Tester...: This should be something to get started pending some clarification of your post. Do you want to add another column or row. And what is to be in that column.

data = Import["https://unstats.un.org/unsd/methodology/m49/", "Data"];
Rest[Part[data,2,1,2,1,2,1][[All,2]]]/. Map[CountryData[#, "UNNumber"] -> # &, CountryData[All]] 

Using the UNCode is problematic as there is an entry in the M49 data that is missing there are ways to fix this but the M49Code is all there. Nevertheless not all number match a Wolfram Language CountryData entity.