Leg bid-price arrays

See the entire bid-price schedule available at the beginning of a sample as an array. This array stores the bid price at each combination of seats sold and days before departure. The tracer averages these opening schedules across retained samples.

This complete array may not be available, depending on the RM system a carrier uses; generally a DP-based algorithm solves for this entire array, while ProBP or SubPB algorithms only solve for the current bid price. You can see this in the example below, where leg 311 (on Brat, using the U RM system) has a complete array, while leg 211 (on Stroop, using the P RM system) only has one column’s worth of data.

To see changes made later in the sample, use the daily vector tracer, which records the vector of active bid prices at each day, instead of the complete array of bid prices at the beginning of the sample. This will reveal a slightly different picture of the bid-price schedule, as the RM system may change the bid prices during the sample. The vector of bid prices actually used 3 days prior to departure (revealed by the daily vector tracer) may not be the same as the expected vector of bid prices 3 days prior to departure when that expectation is computed at the beginning of the sample (this tracer).

See also the other bid-price figures.

Collect the data

Flights 211 and 311 in the Europa UPE demo uses bid prices. Flight 111 does not use bid prices; it can be included in the list of legs, but it will not contribute any observations.

Attach the tracer before running the network:

import passengersim as pax
from passengersim.tracers.bid_price import LegBidPriceArrayTracer

cfg = pax.Config.from_yaml(pax.demo_network("europa/upe"))
cfg.simulation_controls.num_trials = 3
cfg.simulation_controls.num_samples = 300
cfg.simulation_controls.burn_samples = 50
cfg.simulation_controls.show_progress_bar = False
cfg.db = None
cfg.outputs.reports.clear()
cfg.outputs._write_no_files()

with pax.MultiSimulation(cfg) as sim:
    LegBidPriceArrayTracer(leg_ids=[111, 211, 311]).attach(sim)
    summary = sim.run()
table = summary.callback_data["leg_bid_price_arrays"].sort_index(axis=1)
   ╭─────────────────────────────────────╮                                                                         
   │ licensed for jpn                    │                                                                         
   │ until 2036-11-17 20:12:20+00:00 UTC │                                                                         
   │ maximum 42000 legs in network       │                                                                         
   ╰─────────────────────────────────────╯                                                                         

╭───────────────── Summary revenue statistics · Europa ─────────────────╮
│                          Brat            Stroop            Tapas      │
│ Run      Samples   Mean  ~95% CI     Mean  ~95% CI   Mean  ~95% CI    │
│ Europa       750   972K   ±8,871   939.4K   ±9,037   291K   ±2,381    │
│ Post-burn-in only · Approx. 95% CI for mean                           │
╰───────────────────────────────────────────────────────────────────────╯

To enable the same tracer in YAML:

callbacks:
  leg_bid_price_arrays:
    parameters:
      leg_ids: [111, 211, 311]

Read the results

The table has one row per leg and seats-sold level, and one column per day before departure. In the table below, you can see that leg 211 has only one column of data, the rest are missing data.

table.head()
DaysPrior 1 2 3 4 5 6 7 8 9 10 ... 54 55 56 57 58 59 60 61 62 63
leg_id SeatsSold
211 0 312.930092 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
1 315.240991 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
2 317.962503 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
3 320.483404 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN
4 323.090995 NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN

5 rows × 63 columns

The bottom of the table shows leg 311, which has a complete array of bid prices.

table.tail()
DaysPrior 1 2 3 4 5 6 7 8 9 10 ... 54 55 56 57 58 59 60 61 62 63
leg_id SeatsSold
311 146 2.912438 67.537230 183.422365 310.438823 399.801403 456.118702 496.863435 516.986788 529.661174 541.453871 ... 574.204122 574.204122 574.204122 574.204122 574.204122 574.204122 574.204122 574.204122 574.204122 574.204122
147 29.056690 196.585873 334.683284 430.035623 487.678789 527.675217 557.387416 573.661183 585.203219 595.920939 ... 624.614421 624.614421 624.614421 624.614421 624.614421 624.614421 624.614421 624.614421 624.614421 624.614421
148 137.539061 395.294983 514.974548 563.271983 592.400001 615.796686 636.203643 650.556319 662.311468 673.360024 ... 699.398760 699.398760 699.398760 699.398760 699.398760 699.398760 699.398760 699.398760 699.398760 699.398760
149 417.348872 617.270828 680.685734 708.748705 728.821640 746.761958 763.196705 774.806055 784.001044 792.236330 ... 804.114590 804.114590 804.114590 804.114590 804.114590 804.114590 804.114590 804.114590 804.114590 804.114590
150 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 ... 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000

5 rows × 63 columns

Use a leg whose revenue-management system produces bid-price arrays. A leg without an array contributes no observations. To plot the mean prices as a heatmap, choose a collected leg:

from passengersim.tracers.bid_price import fig_leg_bid_price_array

fig_leg_bid_price_array(summary, leg_id=311)

The same figures work for a leg with only one column of data, but the heatmap will be less interesting:

fig_leg_bid_price_array(summary, leg_id=211)

While there won’t be an error from trying to collect a leg without bid prices, the resulting table won’t include it and trying to plot a figure for it will fail.

try:
    fig_leg_bid_price_array(summary, leg_id=111)
except KeyError as e:
    print(f"Leg 111 has no bid-price array: {e} ")
Leg 111 has no bid-price array: nan