Datasets for the Project Portfolio Selection and Scheduling Problem

The primary objective of the Project Portfolio Selection and Scheduling Problem (PPSSP) is to maximise the total portfolio value through the selection and scheduling of a subset of projects subject to various operational constraints. This dataset includes a set of synthetically generated problem instances for two recently-proposed, generalised models of the PPSSP. These problem instances can be used by researchers to compare the performance of heuristic and meta-heuristic solution strategies. In addition, the Python program used to generate the problem instances is supplied, allowing researchers to generate new problem instances.

The first model, extended from [1], is a project-oriented model such that projects are independently selected and scheduled, subject to various operational constraints such as prerequisites and mutual exclusion groups. In the second model, originally proposed in [2], the projects are grouped into predetermined sets, referred to as options. Selection is made on the options, such that all projects within an option must be implemented if the option is selected. However, projects can be independently scheduled, within a fixed window. Furthermore, a project may be present in any number of options, but its cost is only counted once if it is selected as part of multiple options.

Instances in the project_instances folder are divided into three sub-folders (250, 500, 1000) based on the number of projects contained in the problem. Within each sub-folder, instances are named "PI_{index}_{number of projects}_{number of planning years}_{proportion of budget for initiating projects}_{proportion of budget for maintaining ongoing projects}_{time discount rate}.json".

Instances in the option_instances folder are divided into four sub-folders (25, 50, 100, 250) based on the number of families contained in the problem. Within each sub-folder, instances are named "OI_{number of families}_{number of projects}_{proportion of divestment projects}_{yearly budget deviation}_{number of planning years}.json".

[1] K. R. Harrison, S. Elsayed, I. L. Garanovich, T. Weir, M. Galister, S. Boswell, R. Taylor, R. Sarker, A hybrid multi-population approach to the project portfolio selection and scheduling problem for future force design, IEEE Access 9 (2021) 83410–83430.

[2] K. R. Harrison, S. M. Elsayed, I. L. Garanovich, T. Weir, S. G. Boswell, R. A. Sarker, A new model for the project portfolio selection and scheduling problem with defence capability options, in: K. R. Harrison, S. M. Elsayed, I. L. Garanovich, T. Weir, S. G. Boswell, R. A. Sarker (Eds.), Evolutionary and Memetic Computing for Project Portfolio Selection and Scheduling, Springer International Publishing, 2022, pp. 89–123. doi:10.1007/978-3-030-88315-7_5.

项目组合选择与调度问题（Project Portfolio Selection and Scheduling Problem, PPSSP）的核心目标，是在满足各类运营约束的条件下，通过筛选并调度子集项目以最大化组合总价值。本数据集包含两类新近提出的通用PPSSP模型的人工合成算例集，可供研究人员对比启发式与元启发式求解策略的性能。此外，随数据集一并提供了用于生成算例的Python程序，方便研究人员生成自定义算例。

第一个模型拓展自文献[1]，属于面向项目的模型，可独立筛选并调度项目，但需满足前置依赖、互斥组等各类运营约束。第二个模型最早由文献[2]提出，其将项目划分为若干预定义集合，称为“选项（options）”：求解时需对选项进行选择，若选中某一选项，则该选项内的所有项目均需执行，但项目可在固定时间窗口内独立调度。此外，单个项目可隶属于任意数量的选项，若其被多个选项同时选中，仅需计算一次成本。

项目算例文件夹（project_instances）下的算例，依据包含的项目总数划分为250、500、1000三个子文件夹。每个子文件夹内的算例命名规则为："PI_{索引}_{项目数量}_{规划年限数}_{启动项目预算占比}_{存续项目维护预算占比}_{时间折现率}.json"。

选项算例文件夹（option_instances）下的算例，依据包含的选项族数量划分为25、50、100、250四个子文件夹。每个子文件夹内的算例命名规则为："OI_{选项族数量}_{项目总数}_{撤资项目占比}_{年度预算偏差率}_{规划年限数}.json"。

[1] K. R. Harrison, S. Elsayed, I. L. Garanovich, T. Weir, M. Galister, S. Boswell, R. Taylor, R. Sarker, A hybrid multi-population approach to the project portfolio selection and scheduling problem for future force design, IEEE Access 9 (2021) 83410–83430.

[2] K. R. Harrison, S. M. Elsayed, I. L. Garanovich, T. Weir, S. G. Boswell, R. A. Sarker, A new model for the project portfolio selection and scheduling problem with defence capability options, in: K. R. Harrison, S. M. Elsayed, I. L. Garanovich, T. Weir, S. G. Boswell, R. A. Sarker (Eds.), Evolutionary and Memetic Computing for Project Portfolio Selection and Scheduling, Springer International Publishing, 2022, pp. 89–123. doi:10.1007/978-3-030-88315-7_5.