Research on Complex Decision Making Strategies Based on Improved Particle Swarm Optimisation and Machine Learning Techniques

In this paper, we focus on the problem of formulating an optimal decision-making strategy, with special attention to the use of multidimensional data and information, taking into account site distribution characteristics, market risk dynamics and cross-price elasticity between products [1]. Firstly, we propose an innovative decision-making method for the stability problem, i.e., a nonlinear planning model and an improved particle swarm optimisation (PSO) algorithm for solving the problem, which shows an increase in annual return of about 15% [2], [3]. Secondly, to cope with the risk volatility problem, a truncated normal distribution is introduced and dynamically optimised using the PSO algorithm with penalties, and the 7-year prediction results show a significant improvement in the robustness of the system [4]. Finally, considering the substitutability and complementarity between products, a constrained particle swarm optimisation model was constructed and a sensitivity analysis was carried out using the cross-price elasticity and Spearman's correlation coefficient matrix optimisation algorithms, and the optimisation scheme reduced the overall volatility of returns by about 10% [5]. By comprehensively applying advanced optimisation algorithms and machine learning techniques in computer science, this study effectively solves multidimensional problems in complex decision-making, demonstrating the great potential and effectiveness of these techniques in practical applications [6].