Skip to content

Latest commit

 

History

History
138 lines (73 loc) · 4.6 KB

proccessingAmdosimOutput.md

File metadata and controls

138 lines (73 loc) · 4.6 KB
Raw

Proccessing amodsim output data

Output data overview

After you succesfully finish a simulation, files containing information about it's procces are created. Default folder for these is '/your_input_data_folder/experiments/test' but it may be changed in config before the simulation begins (fields amodsim_experiment_dir and experiment_name).

Simulation creates following files:

logs

  • This directory contains standard simulation output log.txt

  • and also logs ??? from gurobi optimizer mip-rebalancing.log

on_demand_vehicle_statistics
  • drop_off.csv - values in format [Simulation Time, Request ID] Each line stands for simulation time [ms], when request with this ID was completed.
  • finish_rebalancing.csv -
  • leave_station.csv -
  • pickup.csv - values in format [Simulation Time, Request ID] Each line stands for simulation time [ms], when request with this ID was started.
  • reach_nearest_station.csv -
  • start_rebalancing.csv -
trip_caches
  • Experiment directory contains one or several trip cache folders, some may be empy.
  • Not neccessary for future use.
allEdgesLoadHistory.json
  • key-value records of how many times has an edge been loaded
darp_times.csv
  • empty??
demand_trip_lengths.csv
  • currently in progress
result.json
  • file containing basic information about executed information such as averageKmWithPassenger, numberOfVehicles, demandsCount etc.
ridesharing.csv
  • Structure is different according to used method of ridesharing (vga or insertion heuristic), variable names are always stated in the head of file
  • HEURISTIC: tuple in format [Batch,New Request Count,Fail Fast Time,Insertion Heuristic Time,Log Fail Time] on each line
    • Batch = number of current iteration (data are not proccessed immediately but in periods, which's lengths are specified in config parameter batch_size)
    • New Request Count - number of requests, that appeared during the batch interval
    • Fail Fast Time -
    • Insertion Heuristic Time -
    • Log Fail Time - Duration of logging the unfeasible requests
  • VGA: tuple in format [Batch,New Request Count, Active Request Count, Group Generation Time, Solver Time, Solver gap], that may be followed by multiple [SIZE Groups count, SIZE Groups Total Time] followed by same number of [SIZE Feasible Groups Count, SIZE Feasible Groups Total Time]
    • Active Request Count - number of waiting requests
    • SIZE Groups count - number of groups of current SIZE, that were created in current batch
    • SIZE Feasible Groups Count - only the groups, which's tasks can be fulfilled by agents
service.csv

  • tuple in format [Demand Time, Demand ID, Vehicle ID, Pickup Time, Dropoff Time, Minimal Possible Service Delay] on each line

    • Demand Time - time of demand creation
    • Minimal Possible Service Delay - Minimal trip duration in ideal conditions
transit.csv
  • Describing times, when vehicle entered 'new edge' / node
  • tuple in format [Transit Time, Static ID, Vehicle State] on each line
    • Transit Time - Simulation time of entering new edge
    • Vehicle state - ordinal od enum OnDemandVehicleState
      • 0 = WAITING
      • 1 = DRIVING_TO_START_LOCATION
      • 2 = DRIVING_TO_TARGET_LOCATION
      • 3 = DRIVING_TO_STATION
      • 4 = REBALANCING
vehicle_occupancy.csv
  • tuple in format [period, vehicle_id, seats_occupied] on each line
    • period is 1 simulation minute

Proccessing data using python

Statistics scripts in python needs several prerequisities. Visualization itself uses python library **matplotlib **, which additionally needs LateX and Ghostscript.

Official matplotlib site regarding this topic: https://matplotlib.org/tutorials/text/usetex.html