Converting the datasubset

FWIW, a!forEach is unexpectedly powerful, and can be nested pretty easily.  This is what I was able to whip up quickly:

a!localVariables(
  
  local!originalData: {
    {impact: 0.0, country_nm: "AA", year: "2019"},
    {impact: 0.0, country_nm: "AA", year: "2020"},
    {impact: 0.0, country_nm: "AA", year: "2021"},
    {impact: 0.0, country_nm: "AA", year: "2022"},
    {impact: 0.0, country_nm: "AA", year: "2023"},
    {impact: 0.0, country_nm: "BB", year: "2019"},
    {impact: 0.0, country_nm: "BB", year: "2020"},
    {impact: 0.0, country_nm: "BB", year: "2021"},
    {impact: 0.0, country_nm: "BB", year: "2022"},
    {impact: 0.0, country_nm: "BB", year: "2023"}
  },
  
  local!uniqueCountries: union(local!originalData.country_nm, local!originalData.country_nm),
  
  a!forEach(
    local!uniqueCountries,
    a!localVariables(
      local!currentCountry: fv!item,
      {
        country_nm: fv!item,
        country_data: a!forEach(
          local!originalData,
          if(
            fv!item.country_nm = local!currentCountry,
            {
              year: fv!item.year,
              impact: fv!item.impact
            },
            {}
          )
        )
      }
    )
  )
)

it seems to do what you want, though in the transformed data set it's necessary to label the "year data" arrays i believe.

Hey ramanh,

Before the solution, I would like to tell you that the dataset you are trying to achieve requires a little modification. You are trying to create a syntactically incorrect dictionary because you mixed the keyword and non-keyword arguments/values. Ex: {country_nm: "BB",{year: "2019", impact:0}}. The first index of the dictionary has "country_nm" as key but the second index does not have any key. The required dataset can be achieved by adding a key to the second index. You can find a code below to achieve the same. 

a!localVariables(
  local!dataSet: {
    {impact: 0.0, country_nm: "AA", year: "2019"},
    {impact: 0.0, country_nm: "AA", year: "2020"},
    {impact: 0.0, country_nm: "AA", year: "2021"},
    {impact: 0.0, country_nm: "AA", year: "2022"},
    {impact: 0.0, country_nm: "AA", year: "2023"},
    {impact: 0.0, country_nm: "BB", year: "2019"},
    {impact: 0.0, country_nm: "BB", year: "2020"},
    {impact: 0.0, country_nm: "BB", year: "2021"},
    {impact: 0.0, country_nm: "BB", year: "2022"},
    {impact: 0.0, country_nm: "BB", year: "2023"}
  },

  a!forEach(
    items: union(local!dataSet.country_nm, local!dataSet.country_nm),
    expression: {
      country_nm: fv!item,
      data:{
        a!forEach(
          items: index(local!dataSet,wherecontains(fv!item,touniformstring(local!dataSet.country_nm)),{}),
          expression: {
            year: fv!item.year,
            impact: fv!item.impact
          }
        )
      }
    }
  )
)