Evolutionary Approaches for Multi-objective Optimization and Pareto-Optimal Solution Selection in Data Analytics

Data analytics analyses complicated healthcare, manufacturing, and other datasets to find insights. It works with real-world limits and can optimize more than one goal at the same time. These are known as multi-objective optimization problems (MOOPs). Numerous MOOP solutions have been suggested for robotic automation, product design, and other applications. This chapter discusses traditional methods such as scalarization, weighted sum, goal programming, and metaheuristics, which are limited to working with a series of single-objective optimization problems or by expanding a Pareto front from a single solution point by point, which may cause the problem to lose its original dimensionality or lead to a suboptimal solution. Nature-inspired algorithms (NIAs), such as evolutionary algorithms, address computational issues using natural phenomenon models and have complicated linkages between numerous dimensions, making them superior at discovering Pareto-optimal MOOP solutions. This chapter provides an overview of evolutionary MOOP, a widely used NIA for solving MOOPs. It also covers concepts derived from differential evolution and genetic algorithms, which utilize a population-based approach. Similar evolutionary methods for Pareto-optimal solution selection are also covered in the chapter. The research covered in this chapter will help readers comprehend evolutionary MOOP issue solving and inspire them to study evolutionary MOOP problems in data analytics.