nate-wessel / routing Goto Github PK

show sets inside parentheses

start from the edge of the time window

Something like:

all_possible_trips = []
for itinerary in best_routes:
   trips = get_from_db()
   all_possible_trips.extend(trips)

207 -> 86
	retro | entropy:0.76 | trips:443
		Pr:0.9, It:{w29,s2615,r43,s2464,w135,s14499,r2,s14525,w69}

...is not nearly as common as...
{w34,s1880,r43,s2280,w135,s14499,r2,s14525,w69}

I need to understand why this is.

Ensure that walk times are treated properly and consistently throughout the code.

on tabbed lines

This is of course impossible. Seen in the all_times.csv output.

> sum(t$any>t$hab,na.rm=T) / nrow(t)
[1] 0.06619754
> sum(t$any>t$real,na.rm=T) / nrow(t)
[1] 0.01286895

It's too big

this may produce unexpected behaviour. I'm not sure why this was removed, but times are not being bounded by a potential walk time.

More can be moved into Itinerary from Path, I think.

But also allow it still to search all possible combinations of points

This will need to go something like:

access(o):
  for d:
    od = OD(o,d)
    od.access(x,y,z)

yadda yadda

Route variation little matters if the time differences are negligible. How can I determine if this is so?

I can export this to CSV and make a quick plot in R

See 12 -> 281

Traceback (most recent call last):
  File "access.py", line 21, in <module>
    access(12)
  File "access.py", line 16, in access
    print( len(od.access()),'trips' )
  File "/home/nate/routing/OD.py", line 150, in access
    trips = db.all_itinerary_trips(self.retro_itin(0))
  File "/home/nate/routing/db.py", line 107, in all_itinerary_trips
    'final_depth':len(itin.segments)
psycopg2.ProgrammingError: cannot subscript type unknown because it is not an array

I believe this has to do with the presence of many suboptimal trips

Make sure all times are handled correctly

Base itin choice on initial wait/walk time + mean time in case of departure time matches.
Also allow for walking option.

This was added to the CSVs to make GTFS files, but would need to be added to the DB directly if I want to use DB based queries.

I'm pulling weekend trips out of the DB, which is not good

Seek to minimize initial wait times.

double check the algorithm and maybe clean up and document