A Novel Discrete Particle Swarm Optimization approach to large-scale survey planning

One of the main challenges in conducting a large-scale face-to-face survey is the efficient planning and deployment of manpower resources to perform the interviews in an orderly manner to support productivity. In this paper, a swarm-based algorithm is proposed to autonomously perform the assignment of respondents to interviewers to facilitate the fieldwork planning of a face-to-face survey involving close to 32,000 pre-selected respondents and 100 interviewers. The assignment is to be performed in two stages. Firstly, K-means clustering is used to group/cluster the survey respondents based on their geographical locations. This ensures that respondents staying close to one another are visited by the same interviewer in the same survey period. The second stage employs a novel Selective Probabilistic Discrete Particle Swarm Optimization (SPD-PSO) technique to assign the respective clusters of respondents to the interviewers to minimize the travel distance of each interviewer while trying to achieve a near-even spread of the survey workload across all the interviewers. The effectiveness of the proposed SPD-PSO algorithm is then evaluated by comparing its allocation results against those of two selected benchmark methods.