Multi-agent Kidney Exchange Program: dataset for simulation along time horizon

The dataset contains the instances files for the paper X. Klimentova, A. Viana, J. P. Pedroso, N. Santos. Fairness models for multi-agent kidney exchange programmes. To appear in Omega: The International Journal of Management Science (2020). The same dataset was also use in Monteiro, T., Klimentova, X., Pedroso, J.P.Pedroso, A. Viana. A comparison of matching algorithms for kidney exchange programs addressing waiting time. Cent Eur J Oper Res (2020). https://doi.org/10.1007/s10100-020-00680-y

Each instance mimics pools of kidney exchange progammes of several agents (e.g. countires) over time. Incompatible donor-recipient pairs appear and leave along the time horizon. Each of the pairs belongs to the pool of one of the agents. The virutal compatiblity among pairs is represented on a directed graph G = (V,A), called compatibility graph, where the set of vertices V corresponds to the set of incompatible pairs and non-directed donors. An arc from a vertex i to a vertex j indicates compatibility of donor in i with the patient in j. The positive real crossmatch testing is also incorporated by saving the arcs that would fail in case they are chosen is a cycle in one of the matching runs. The generator creates randomly graphs based on probabilities of blood type and of donor–patient tissue compatibility; the arrival of pairs and non-directed donors is generated based on a given arrival rates. An instance of the dataset represents a pools of 4 agents, that are simulated for the period of 6 years.

There are 100 instances compressed in 4 zip-archives, each containing 25 instances. Each of the instances is described by 3 files, where index s is the seed used for random function when generating the instance.

a) characterisations_s.csv -- csv file that contains information on each pair in the merged pool in the following columns

0 :  Pair ID

1 : Donor ID

2 : Donor blood type

3 : Donor age

4 : Patient ID

5 : Patient blood type

6 : Patient PRA

7 : Patient cross-match probability

8 : Patient age

9 : Pair arrival day

10 : Pair departure day

11 : Pair probability of failure

12 : Pair from pool (e.g. country to which the pair belongs to)

In case of non-directed donor the information about the patient is filled by -1;

b) acrs_s.csv - csv file containts the compatibility graph of the problem described above. In the first line the file contains values n – number of vertices in the graph and m – number of arcs in the graph. In the following m lines of the file, the existing arcs (i,j) are presented as follows: i j w_ij where i and j are IDs of pairs, w_ij is the weight of the arc, which is always equal to 1.0 for all the instances in this dataset.

c) fail_arcs_s.csv - is the list of arcs that would fail due to positive crossmatch test in case they appear in a chosen cycle or chain in any matching run. The format of the file is the same as that for arcs_s.csv. The first line represents the n - number of vertices in the graph, and m_fail the number of failed arcs listed in the following m_fail lines in the same way as in arcs_s.csv