States which appeared in the NA data and had 2000 or fewer GTAs in the RB data - i.e. states that we may want to target scraping for.
## [1] "ES" "PB" "PI" "PR" "RJ" "RN" "SE"
Also a visual assessment shows that in MG & PE we had a lot of GTAs from NA in earlier years, though this has now dropped off - these would be worth targeted scraping.
Below I list these states in order of what proportion of exports they made up in 2015-2017. Only four states exported.
## # A tibble: 4 x 2
## state PERC_VOL
## <chr> <dbl>
## 1 MINAS GERAIS 7.62
## 2 PARANA 1.78
## 3 ESPIRITO SANTO 0.34
## 4 RIO DE JANEIRO 0.19