Consumer portfolio management with usage of evolutionary approach by PSO algorithm and machine learning models

izvorni znanstveni rad

izvorni znanstveni rad

Consumer portfolio management with usage of evolutionary approach by PSO algorithm and machine learning models

Vrsta prilog u knjizi
Tip izvorni znanstveni rad
Godina 2024
Nadređena publikacija Recent Trends in Swarm Intelligence Enabled Research for Engineering Applications
Stranice str. 187-211
DOI 10.1016/b978-0-443-15533-8.00010-2
Status objavljeno

Sažetak

The chapter presents novel approaches in the usage of particle swarm optimization (PSO) algorithms together with techniques like Bayesian networks and text mining analytics in the service of consumer portfolio management. The optimal solution will cover situations where complexity is present according to numerous factors that directly affect customer satisfaction. Customer satisfaction is a complex topic that cannot be observed one-dimensionally, and it unites many different business aspects. We can assume that service quality level has an impact on customer satisfaction, and we can produce one predictive model that can evaluate satisfaction level, but there are many other aspects like post-buying support, recommendation processes, and other dimensions that contribute to customer satisfaction. Each of the mentioned areas can be covered by separate machine learning models, and at the end each machine learning model has a significant complexity level due to numerous input attributes and outputs forming separate models.
To optimize customer satisfaction for different individual cases, all the separate machine learning models must be united into one solution. That will increase the level of complexity, and we see the PSO algorithm as an integrator and solution seeker in those conditions. Consumer portfolio management in this case means optimization of offers, recognition of customer satisfaction level, and having recommended actions for mitigation of negative trends within the portfolio. The chapter proposes a holistic solution that includes the early warning system development-based elements discussed.

Ključne riječi

AI, Evolutionary computing, PSO, machine learning