svraka/teroszt

🗺🇭🇺 R package with data for Hungarian administrative and statistical divisions

teroszt's Issues

Add support for alternative geographic visualization techniques

This introduction by Ryan Hafen offers some possibilities, including the geofacet package.

Replace OSM maps

Data downloaded from OpenStreetMap is outdated. It looks like they are still using the original 2013 version and district boundaries were redrawn several times. The whole point of including OSM data was to have district level maps, so this needs to be fixed.

A better approach would be to use Eurostat's regularly updated GISCO data. Download settlement maps and and (using giscoR) and aggragate it to district level using sf. As a proof of concept, something like this could be bundled into a function:

lau_hu <- gisco_get_lau(country = "hu")

sf_use_s2(TRUE)
jaras_sf <- lau_hu %>%
  left_join(select(tsz_2018, torzsszam, jaras, jaras_nev),
            by = c("LAU_CODE" = "torzsszam")) %>%
  group_by(jaras, jaras_nev) %>%
  summarise()

Although this needs some further cleaning, as there are some strange districts, e.g. Kaszaper (in the Mezőkovácsháza District) has a borough called Pusztaszőlős, which is an exclave in the Orosháza District.

This would also fix R CMD check notes about package size.

Improve documentation

The datasets are well documented but at least the README should have some examples and plots, especially maps. A vignette would be too much at this point.

Add territorial IDs to map data

kozighatarok_2018 can only be joined to other data using names (NAME column). IDs are probably safer, and are available for ADMIN_LEVE values, except for Budapest neighbourhoods.

Check new data releases

Currently the latest data is for 2018. Check if there were any changes.

Set up GitHub Actions

The R community moved away from Travis.

Should this packages have helper functions?

There some recurring tasks, which could be automated in functions, e.g. get a data frame of county names and codes:

tsz_2018 %>% 
  distinct(megye_nev, megye)

This would require moving dplyr to Imports.

Add postal codes for post box addresses

Post box addresses and similar technical postal codes can be found in many databases but unfortunately Magyar Posta's database does list them and the list of post offices (xlsx) is incomplete. Try to ask Posta for them, or manually compile a list from postal codes found in real datasets.

Better support for working with postal codes

As postal codes can cross county and district boundaries, postal code based classifications require significant data cleaning. Ideally there should be crosswalk tables where overlapping postal codes are manually cleaned and validated. The following crosswalks are relevant:

• Regions: very few overlapping postal codes
• Counties: more but easily manageable
• Districts: even more, needs heavy manual cleaning.
• Metropolitan areas: also a lot of manual cleaning

Use factors in label names

Label type columns should be factors to keep their commonly used ordering based on their coding system. E.g. regions are usually sorted based on NUTS codes, not alphabetically. This is relevant for tsz_2018 and nav_igazgatosagi_kodok.

Add English translations to settlement types

jogallas_2005_nev in tsz_2018.