Basic stuff: relating addresses from the tickets to Minsk city districts via checking them through publicly available street lists.
This is a mini-project of a big task related to analyzing publicly available data of municipal issue tracker (so to speak) 115∙BEL, which took place at Open Data Laboratory event at 20th of May 2017 (Minsk, OpenDataBY).
Original dataset is from OpenData.BY.
- good.txt – attributed data (tab-separated)
- ambig.txt – items which are ambiguous, because either they don't have a type of object (street or avenue, kind of), and there are several objects with the name like this, or the street has building related to different city districts and I don't have enough data to attribute it to any of them.
- notfound.txt – stuff that failed to be attributed and there is no guess, mostly the reason is it's not in Russian. There is a need to process Belarusian as well, but sources I used are Russian-only (which is, by the way, very bad, because one of the sources is the site of local administration – and it's neither actual, nor bilingual).
- notparsed.txt – stuff is mostly like previous one, but even more differs from the content of sources.
- data.csv – list of processing of good.txt, formatted like this: “city district” – “amount of tickets”.
I'm not happy about data sources: they have a lot of evident mistakes and don't contain actual data on every street, despite second site is an official source (or must be?).
Currently we have 63543 tickets, 60408 were attributed to district, other failed: either they are in Belarusian, or in transliteration, or contain typos, or their addresses are really complicated non-machine-readable descriptions of some places.
I made simple visualization of the data in D3.js. There is a screenshot:
