d610f20 / random-activitygen-py Goto Github PK

This seems like a pattern that real cities follow. See image of Aalborg, blue dots are generated schools, red squares are real, validated schools.

Apparently the display function assumes that the width and height are equal, when they're not, this happens:
When x=1500, y=1000:

When x=1000, y=1500:

Related to #40 and depends on #39

Use a non-linear function to scale school dots in the render, so that the dots will not cover too much of the image

As found in #9, a user may want or may not want existing streets to be overwritten with population and industry values from noisemap.

resolved. wrong cmd command

Perlin noise takes seed, scale, octaves, lucanarity, and persistence arguments. When testing or visualisation facilities are available they should be investigated as to which settings generate more realistic noisemaps.

This is a FIXME in logging_setup() e.g. allowing stdout to print in INFO and file to output in DEBUG.
Ref: https://stackoverflow.com/questions/25187083/python-logging-to-multiple-handlers-at-different-log-levels

If image size is very large, dots are too small, and cannot be seen unless zooming in

The render only creates straight roads, meaning that busstops/schools looks like they are placed outside of edges

Running time python randomActivityGen.py -n in/aalborg_metro.net.xml -s in/aalborg_metro.stat.xml -o /out/aalborg_metro.stat.xml --verbose --display on OSM downloaded map of Aalborg metropolitan area.

Traceback (most recent call last):
  File "randomActivityGen.py", line 160, in <module>
    main()
  File "randomActivityGen.py", line 147, in main
    setup_bus_stops(net, stats, int(args["--bus-stop.distance"]), int(args["--bus-stop.k"]))
  File "randomActivityGen.py", line 92, in setup_bus_stops
    for i, busstop in enumerate(bus_stop_generator(edges, min_distance, min_distance*2, k, seeds=seed_bus_stops)):
  File "/home/user/projects/random-activitygen-py/bus.py", line 80, in bus_stop_generator
    point = next(filter(
  File "/home/user/projects/random-activitygen-py/utility.py", line 187, in firstn
    yield next(gen)
  File "/home/user/projects/random-activitygen-py/bus.py", line 24, in _road_point_generator
    pos = position_on_edge(road, remaining)
  File "/home/user/projects/random-activitygen-py/utility.py", line 126, in position_on_edge
    for coord1, coord2 in (edge.getShape()[i:i + 2] for i in range(0, int(len(edge.getShape())), 2)):
ValueError: not enough values to unpack (expected 2, got 1)

Maybe allow user to choose to save the image or not.
Save only the latest image overwriting previous iamges, perhaps name it netFile-statsFile so that it is easy to identify the files used to generate it.

Useful for users when they want to view a previously generated statistics file applied on a network.
Useful for developers when working with the display mechanism.

City gates are not overwritten, schools should be either.

For some reason the single school in Vejen is always placed on the same edge regardless of noise seed.

Instead of putting logs like the following:
logging.debug(f"Using begin_age: {begin_age}, end_age: {end_age} for school(s)")
We should categorise them, so it is easy to read through a long logfile and understand the context of each event:
logging.debug(f"[school] Using begin_age: {begin_age}, end_age: {end_age} for school(s)")
The above is just an idea but seems to work okay.

We currently assume that the user has a half-finished stats file. It case they don't we can probably automatically guestimate the number of inhabitants and number household etc. from the network's size. Other values can just be some default values to generate an average city.

Typically, you'd expect perlin noisemaps to look like this: (with a base=0)

But with a base=-88 this pattern is had:

.