Script to generate a sankey graph of the brazilian government revenue

generate_sankey_graph_fed_gov_revenue.r

devtools::install_github("tchiluanda/rsiconfi")
library(rsiconfi)
library(networkD3)
library(tidyverse)


#Get data in SICONFI for budget revenue (Ï-C) related to the federal government entity (1) in 2018
df_rec <- rsiconfi::get_dca(2018,"I-C","1")

# find hierachical strtucture of budget revenue
df_rec_trabalho<-
  df_rec %>% 
  #filter(stringr::str_starts(cod_conta,"R")) %>%
  filter(coluna == "Receitas Brutas Realizadas") %>%
  mutate(nivel_0 = cod_conta =="TotalReceitas" ) %>%
  mutate(nivel_1= (stringr::str_ends(cod_conta,".0.0.0.00.0.0"))) %>%
  mutate(nivel_2= ((stringr::str_sub(cod_conta,5,5)!= "0") & (stringr::str_ends(cod_conta,"0.0.00.0.0")))) %>%
  mutate(nivel_3= ((stringr::str_sub(cod_conta,7,7)!= "0") & (stringr::str_ends(cod_conta,"0.00.0.0")))) %>%
  filter(nivel_0 | nivel_1 | nivel_2 | nivel_3) %>%
  mutate(pai = case_when(
    nivel_1 ~ "TotalReceitas",
    nivel_2 ~ paste0(stringr::str_sub(cod_conta,1,3), ".0.0.0.00.0.0"),
    nivel_3 ~ paste0(stringr::str_sub(cod_conta,1,5), ".0.0.00.0.0")
  )) %>%
  mutate (source = row_number() -1)


v_pai <- 
  df_rec_trabalho %>%
  mutate(pai = ifelse(is.na(pai),cod_conta,pai))


pos_pai<-  
  map_int(v_pai$pai, function(a_pai){
    print(a_pai)
    which(df_rec_trabalho$cod_conta==a_pai)
    
  })


pos_pai<- pos_pai -1

df_rec_trabalho$destination <- pos_pai

#identify the nodes for sankey graph
nodes<- df_rec_trabalho %>%
  mutate(conta= ifelse(cod_conta == "TotalReceitas","Total de Receitas Brutas", stringr::str_sub(conta, 18,str_length(conta)))) %>%
  select(conta)


#identify the links for sankey graph
links<- 
  df_rec_trabalho %>%
  filter(!is.na(pai)) %>%
  select(source,
         destination,
         valor)

#generate sankey graph
#use viewer zoom if you are using RStudio
sankeyNetwork(Links = links, Nodes = nodes, Source = "source",
              Target = "destination", Value = "valor", NodeID = "conta",
              units = "", fontSize = 12, nodeWidth = 30)