Blogger diamond geezer recently calculated London's most average bus stop, geographically speaking using data from a FoI response from TfL.
But TfL aren't the only people who list all the bus stops. There's also National Public Transport Access Nodes (NaPTAN), a UK-wide list of all transit stops.
I run buildmorebuslanes.com (code), so I happen to have this information in a database table:
travelinedata=> with
by_lat as (
select latitude, longitude, name, rank() over (order by latitude) from naptan
),
by_lon as (
select latitude, longitude, name, rank() over (order by longitude) from naptan
)
select 'lat', latitude as value, name, latitude, longitude
from by_lat
where rank = (select count(*) from naptan)::int/2
union
select 'lon', longitude as value, name, latitude, longitude
from by_lon
where rank = (select count(*) from naptan)::int/2
order by 1;
?column? | value | name | latitude | longitude
----------+----------+-------------------------+----------+-----------
lat | 52.606 | Lancaster School | 52.606 | -1.12355
lat | 52.606 | Moat House Lane East | 52.606 | -2.06927
lon | -1.76065 | Somerset Road Broadgate | 53.6405 | -1.76065
(3 rows)
The location of this is a field near Bangley Lane, Lichfield.
Using these co-ordinates, the nearest bus stop is:
travelinedata=> select * from naptan order by abs(-1.76065 - longitude) + abs(52.606 - latitude) asc limit 3;
atcocode_id | code | name | latitude | longitude
-------------+----------+---------------------+----------+-----------
221103 | | Gainsborough Avenue | 52.6136 | -1.74121
223852 | stagmtdt | Gainsborough Drive | 52.6136 | -1.74113
221102 | | Gainsborough Drive | 52.6146 | -1.7402
(3 rows)
It looks like Gainsborough Avenue doesn't really exist, so this is the nearest one really Gainsborough Drive, Mile Oak
We took the median lattitude and median longitude.
But there's no good reason for that: we could have picked any set of orthogonal axes.
What if we rotated our axes by 45° (τ/8)? Would that affect our result?
Let's define some rotated axes:
slantitude = cos(τ/8)*latitude + sin(τ/8)*longitude = (latitude + longitude) / sqrt(2)songitude = cos(-τ/8)*latitude + sin(-τ/8)*longitude = (latitude - longitude) / sqrt(2)The sqrt(2) scaling factors don't matter, so let's do some queries:
travelinedata=> with
naptan_transformed as (
select
latitude + longitude as slantitude,
latitude - longitude as songitude,
name
from naptan
),
by_slant as (
select slantitude, songitude, name, rank() over (order by slantitude) from naptan_transformed
),
by_song as (
select slantitude, songitude, name, rank() over (order by songitude) from naptan_transformed
)
select 'slant', slantitude as value, name, slantitude, songitude
from by_slant
where rank = (select count(*) from naptan_transformed)::int/2
union
select 'song', songitude as value, name, slantitude, songitude
from by_song
where rank = (select count(*) from naptan_transformed)::int/2
order by 1;
?column? | value | name | slantitude | songitude
----------+---------+-----------------+------------+-----------
slant | 51.2967 | Belsize Road | 51.2967 | 52.2001
song | 54.5401 | Masons Cottages | 51.9507 | 54.5401
(2 rows)
travelinedata=> with
naptan_transformed as (
select
latitude + longitude as slantitude,
latitude - longitude as songitude,
*
from naptan
)
select * from naptan_transformed order by abs(51.2967 - slantitude) + abs(54.5401 - songitude) asc limit 5;
slantitude | songitude | atcocode_id | code | name | latitude | longitude
------------+-----------+-------------+----------+---------------+----------+-----------
51.2914 | 54.5385 | 43790 | dbsadpjp | Trusley Manor | 52.915 | -1.62355
51.2974 | 54.5629 | 43776 | dbsgtadw | Osleston Hall | 52.9301 | -1.63273
51.2968 | 54.5149 | 43754 | dbsgwpdt | Main Street | 52.9059 | -1.60905
51.3239 | 54.5395 | 46166 | dbsgdatd | Black Cow | 52.9317 | -1.60779
51.313 | 54.5655 | 47772 | dbsdtjdg | Church | 52.9393 | -1.62627
(5 rows)
Ignoring bus stops which aren't in use (I think?), the closest one is Long Lane village church in Derbyshire.
This is about 50km away - about the diamater of London! Wowzers - the method has a huge impact!
Long Lane village has one bus per day (below the threshold for the default map view).
Meanwhile, in Mile Oak, Lichfield, there are 4 buses an hour going by, traveling at over 27mph. That's a very good service for a small village.
That service is the Arriva Midlands service 110, going Tamworth to Birmingham.
The network service map suggests this is for commuters in Sutton Coldfield going to Tamworth. Mile Oak just happens to be on the road between them.
The difference between 1 bus a day or 50 buses a day? It really pays to "be on the way", as Jarett Walker would put it.