This readme file provides guidance on how to interpret the data in the files "Data_for_Tables_1-5.csv", "Additional_data_for_Table_5.csv" and "Data_for_Table_6.csv".
The data values in the various columns of "Data_for_Tables_1-5.csv" are described below. Please refer to Appendix A.8 in the supplementary material of our paper for further explanation about the various parameters. Note that some of the parameters (e.g. Lambda, phi) are required for some scenarios but not for others, depending on the nature of the demand process. If a parameter is not required in a particular scenario, its value is missing in the relevant row of the dataset.
Column index (A-Z) Parameter/Output
================== ================
A The scenario identifier (from 1 to 28800)
B The type of demand process (Poisson or PSRA)
C The demand sub-case (explained in Appendix A.8)
D No. of linear constraints on OTE
E alpha (weighting parameter in objective function)
F beta (weighting parameter in objective function)
G k (Erlang distribution parameter)
H tau (time interval duration)
I S (no. of time intervals)
J delta_{OTE} (OTE variance parameter)
K delta_{dem} (demand process parameter)
L rho_{OTE} (used to constract the shape of the OTE)
M Lambda (used as a demand process parameter)
N phi (also used as a demand process parameter)
O xi (delay parameter for PSRA demand process)
P x0 (initial phase count for class A customers)
Q y0 (initial phase count for class D customers)
S normalised total cost for the MDR heuristic (relative to MDR)
T normalised total cost for the MYP heuristic (relative to MDR)
U normalised total cost for the API heuristic (relative to MDR)
V normalised total cost for the CAV heuristic (relative to MDR)
W normalised total cost for the LSF heuristic (relative to MDR)
The data values in the various columns of "Additional_data_for_Table_5.csv" are described below.
Column index (A-Z) Parameter/Output
================== ================
A Scenario index in the reduced sample of 4800 scenarios
B p (parameter used to re-scale demand rates)
C percentage cost difference between MDR and MYP
D percentage cost difference between MDR and API
E percentage cost difference between MDR and CAV
F percentage cost difference between MDR and LSF
The data values in the various columns of "Data_for_Table_6.csv" are described below. This file mainly follows the same format as "Data_for_Tables_1-5.csv", except the parameters alpha and beta are omitted because these are equal to 1 and 0 (respectively) in all scenarios.
Column index (A-Z) Parameter/Output
================== ================
A The scenario identifier (from 1 to 4800)
B The type of demand process (Poisson or PSRA)
C The demand sub-case (explained in Appendix A.8)
D No. of linear constraints on OTE
E k (Erlang distribution parameter)
F tau (time interval duration)
G S (no. of time intervals)
H delta_{OTE} (OTE variance parameter)
I delta_{dem} (demand process parameter)
J rho_{OTE} (used to constract the shape of the OTE)
K Lambda (used as a demand process parameter)
L phi (also used as a demand process parameter)
M xi (delay parameter for PSRA demand process)
N x0 (initial phase count for class A customers)
O y0 (initial phase count for class D customers)
Q normalised average waiting time for the MYP heuristic (relative to MYP)
R normalised average waiting time for the API heuristic (relative to MYP)
S normalised average waiting time for the CAV heuristic (relative to MYP)
T normalised average waiting time for the LSF heuristic (relative to MYP)