Manipulating Skim matrices

In this example we show how to open and manipulate skim matrices

Imports

from pathlib import Path

import numpy as np
from polaris import Polaris

Data Sources

Open the demand database for analysis

model_fldr = "/tmp/Grid"
project = Polaris.from_dir(Path(model_fldr))

highway_skim = project.skims.highway
pt_skim = project.skims.transit

We can see what intervals exist in a matrix

print(highway_skim.intervals)

[240, 360, 420, 480, 540, 600, 720, 840, 900, 960, 1020, 1080, 1140, 1200, 1320, 1440]

For accessing costs, distances and time can be done in two different ways for highway skims

time240 = highway_skim.time[240]
distance240 = highway_skim.distance[240]
cost240 = highway_skim.cost[240]
dist_skim = highway_skim.get_skims(interval=480, metric="distance")

For PT skims there is only one way

time_pt_240 = pt_skim.get_skims(interval=240, mode="Bus", metric="time")

We can also plot a histogram (Distance frequency distribution)

import matplotlib.pyplot as plt

c, x = np.histogram(np.ravel(dist_skim), bins=50, density=True)
plt.plot(x[:-1], c)
plt.title("Distribution of auto distance skim for 8am")
plt.xlabel("Distance (m)")

Text(0.5, 23.52222222222222, 'Distance (m)')

For PT skims we can list modes and metrics as well

print(pt_skim.intervals)
print(pt_skim.modes)
print(pt_skim.metrics)

[240, 360, 420, 480, 540, 600, 720, 840, 900, 960, 1020, 1080, 1140, 1200, 1320, 1440]
['BUS', 'RAIL', 'PARK_AND_RIDE', 'PARK_AND_RAIL', 'RIDE_AND_UNPARK', 'TNC_AND_RIDE']
['time', 'walk_access_time', 'auto_access_time', 'wait_time', 'transfers', 'fare']

We can remove intervals for the skim matrix

highway_skim.remove_interval(1020)
pt_skim.remove_interval(1020)

We can also add new skims to the file

First with ZEROS

# For highway skims
highway_skim.add_interval(1000)
print("Highway intervals", highway_skim.intervals)
print("Highway matrix sum", highway_skim.time[1000].sum())

# For PT skims
pt_skim.add_interval(1010)
print("Transit intervals", pt_skim.intervals)
print("Transit matrix sum", pt_skim.get_skims(interval=1010, mode="Bus", metric="time").sum())

Highway intervals [240, 360, 420, 480, 540, 600, 720, 840, 900, 960, 1000, 1080, 1140, 1200, 1320, 1440]
Highway matrix sum 0.0
Transit intervals [240, 360, 420, 480, 540, 600, 720, 840, 900, 960, 1010, 1080, 1140, 1200, 1320, 1440]
Transit matrix sum 0.0

Or by copying another interval

# For Highways
highway_skim.add_interval(1350, copy_interval=1320)
print("Highway intervals", highway_skim.intervals)
print("The time skim totals match:", highway_skim.time[1320].sum(), highway_skim.time[1350].sum())
print("The distance skim totals also match:", highway_skim.distance[1320].sum(), highway_skim.distance[1350].sum())
print("The cost skim totals also match:", highway_skim.cost[1320].sum(), highway_skim.cost[1350].sum())

# For PT
pt_skim.add_interval(1390, copy_interval=1320)
print("Transit intervals", pt_skim.intervals)
print(
    "The time skim totals match:",
    np.nansum(pt_skim.get_skims(interval=1390, mode="Bus", metric="time")),
    np.nansum(pt_skim.get_skims(interval=1320, mode="Bus", metric="time")),
)

Highway intervals [240, 360, 420, 480, 540, 600, 720, 840, 900, 960, 1000, 1080, 1140, 1200, 1320, 1350, 1440]
The time skim totals match: 1574.3708 1574.3708
The distance skim totals also match: 1596717.8 1596717.8
The cost skim totals also match: 0.0 0.0
Transit intervals [240, 360, 420, 480, 540, 600, 720, 840, 900, 960, 1010, 1080, 1140, 1200, 1320, 1390, 1440]
The time skim totals match: inf inf

Or by copying another interval Since all the changes we made to the matrices only live in memory, we write them down to disk

hwy_skim_file = Path("/tmp/new_highway_skim.omx")
pt_skim_file = Path("/tmp/new_pt_skim.omx")
hwy_skim_file.unlink(missing_ok=True)
pt_skim_file.unlink(missing_ok=True)
pt_skim.write_to_file(pt_skim_file)
highway_skim.write_to_file(hwy_skim_file)