Data and sources

Facility data

Facilities come from two sources, and each category uses whichever of them actually holds the places. Both are imported in full once a month and served from our own database. Building a report queries nothing outside this site.

• OpenStreetMap — stations, bus stops, convenience stores, supermarkets, police boxes, post offices, parks, sights, public baths and laundromats. © OpenStreetMap contributors, under the Open Database License (ODbL), imported from the Geofabrik extract of Japan.
• Overture Maps Foundation — restaurants, cafes, bars, pharmacies, ATMs, coin lockers, souvenir shops and currency exchange. © Overture Maps Foundation, under CDLA-Permissive 2.0 (parts derived from Foursquare are under the Apache License 2.0).

The split is measured, not preferred. OpenStreetMap holds 134,350 restaurants for the whole of Japan, against roughly 490,000 in the national census — and the gap is widest away from the big cities: within 800 m of one regional station it held five, where Overture holds 54, each of which we checked by hand. Police boxes and stations go the other way, so those stay with OpenStreetMap.

The map behind the page

The map you can see behind these pages is the same OpenStreetMap data. Japan is cut out of the Protomaps daily build as a single PMTiles archive, kept in our own storage and drawn with MapLibre GL. © OpenStreetMap contributors, under the ODbL. No tiles are fetched from anybody else's service, and nothing about it is billed per view.

Search

Stays, stations, neighbourhoods and landmarks all come from the database above. Searching queries nothing outside this site.

How the score is built

For each category, every matching place within 800 m counts towards a target. Anything within 150 m counts in full; credit falls off in a straight line to zero at 800 m. Category scores are combined with fixed weights that add up to 100.

Walking times use 80 metres per minute. That is the convention Japanese property listings are required to use; it is not an international standard, which is why the measured distance in metres is always printed next to it.

What is deliberately left out

Flood and landslide hazard zones, road and railway noise, and earthquake and tsunami risk are not part of the score. They are listed on every report as excluded. A category that is missing from the data is never presented as “nothing found there”.

Laundromats (4,329 nationwide), souvenir shops (2,992), currency exchange (197) and diving shops and marinas (898) are shown but not scored: few enough that a low count means nobody recorded them rather than that there are none. Marking a place down for that would say something false about it. A hit is worth showing; a miss is not evidence.

This judgement is revisited whenever the sources change. Coin lockers and ATMs were excluded for the same reason — 343 and 6,085 of them at the time. Adding a source that actually holds them took those figures to 15,151 and 61,548, and both now count towards the score. What counts as an ATM is narrower still: the 74,676 that take a card issued outside Japan, since most Japanese bank machines decline them. Most Japan Post machines sit inside a post office, so post offices (24,430) are counted here too — except the sub-post offices (3,799), which often have no machine and are listed as post offices without being counted as cash points. Whether a category is scored depends on whether the data represents reality, and the answer is revised every time we measure again.

Accuracy

Distances are straight-line, not walking routes, so a river or a railway between you and a shop is not accounted for. Both sources can be out of date — OpenStreetMap is contributor-maintained, and neither source tells us when a business has closed. Check anything that matters to your trip before you rely on it.