Reshape long

New to Julia
, ,
5

Posting an attempt of a more general function, though it’s much slower than Douglass:

function reshape_long(df, id, cons, stubs)

  stub_len = length(stubs);
  df_vector = [DataFrame() for _ in 1:stub_len];

  iter = 1
  for i in 1:length(df_vector)
    if iter ==1

      df_vector[i] = select(df, [id, cons], Regex(stubs[i]))
      iter += 1

    else
      df_vector[i] = select(df, [id], Regex(stubs[i]))

    end
  end

  df_vector2 = [DataFrame() for _ in 1:stub_len];

  for i in 1:length(df_vector2)
    df_vector2[i] =  stack(df_vector[i], variable_name = stubs[i], Regex(stubs[i]),  value_name = Symbol("value_$i") );

    df_vector2[i] = combine(groupby(df_vector2[i], id), sdf -> sort(sdf), id => eachindex => :index)
  end



  res = [DataFrame() for _ in 1:(stub_len-1)];


  iter = 1
  for i in 1:length(res)

    if iter ==1
      res[i] = outerjoin(df_vector2[i],df_vector2[i+1], on = [id,:index] )
      iter +=1
    else
      res[i] = outerjoin(res[1], df_vector2[i+1], on = [id, :index])
    end


  end

  d = res[end]
  symbs = Symbol(stubs[1])
  d[!, symbs] = getproperty.(match.(r"[0-9]+",d[!, symbs]), :match)


  rename!(d,symbs => :year)
  select!(d, Not([ Symbol(stubs[i]) for i in 2:length(stubs) ]))
  rename!(d, [(Symbol("value_$i")=>Symbol(stubs[i])) for i in 1:length(stubs)])
  select!(d, Not(:index, ))

  return(d)

end

Example: reshape_long(df, :ID, :sex, ["inc", "ue"])