ORiON

Editorial to Volume 38(2)

Jaco Visagie — 2023-02-10

A mathematical model to select the optimal age group classified freestyle relay teams for a masters swimming competition

Inger Fabris-Rotelli — 2022-12-22

This study extends the research frontier of relay swim team selection. In particular, the
extension is to select the optimum teams for a club competing at a masters competition. The
model uses an integer optimisation routine to select several teams of different cumulative
ages who fall into the age classification categories defined at masters level. Data analysed to
demonstrate the method is based on the best times of masters competitors for a swimming
club in South Africa.

Estimating the dependence parameter in bivariate extreme value statistics through a Bayesian approach

Andrehette Verster — 2022-12-29

Quantitatively modelling opinion dynamics during elections

Michael Burke — 2022-12-29

Political advertising has become overwhelmingly focused on online platforms, with ever-improving capabilities of tailored and targeted advertising to individuals. Modelling the effect of political propaganda on a voting population has become more prevalent in opinion dynamics research, which has become further enabled by the applications of computer simulation and data analysis.
In this study we explore the effect of propaganda on a voting population. The agent-based model describes a population of voter agents who hold a political opinion using knowledge and emotion as their control variables. These variables are updated through agent interactions and political propaganda. The model is based on the emotion/information/opinion (E/I/O) approach which is applied to a grid-based and network population. Furthermore, the network is programmed to be partially dynamic, in that connections between disagreeing agents can be severed under certain conditions concerning the intimacy of agent relationships. This is performed with the intention of adding more flexibility to the model, whilst making it a more realistic representation of reality. The different network types are shown to produce varying proportions of metastable agent states from identical starting conditions, which can be used to represent real political situations and predict future change.

Towards a framework for predicting packing algorithm performance across instance space

Rosephine Georgina Rakotonirainy — 2022-12-22

Finding the conditions under which packing algorithms succeed or fail with respect to a set of test instances is crucial for understanding their strengths and weaknesses, and for automated packing algorithm selection. This paper tackles the important task of objective packing algorithm selection. A framework for understanding the relationship between critical features of packing problem instances and the performance of packing algorithms is proposed. The framework can be used to predict algorithm performance on previously unseen instances with high accuracy and can be applied to find predictions in other instances of cutting and packing problems. It can also be applied to determine the relative strengths and weaknesses of each algorithm within the instance space. The effectiveness of the framework is demonstrated using the two-dimensional strip packing problem as a case study.

Methods of enhancing the MOO CEM algorithm

James Bekker — 2022-12-22

With the increasing need to solve problems faster and with fewer resources, great emphasis
is placed on optimisation. Many real-world problems require addressing more than one
objective that are in conflict, as well as taking into consideration a number of practical restrictions
or constraints. The multi-objective optimisation using the cross-entropy method (MOO
CEM) algorithm is one of many algorithms that addresses the need to solve multi-objective
problems effectively, but it has a number of limitations. This paper explores methods of
enhancing the MOO CEM algorithm in order to improve the efficiency and increase the functionality
of the algorithm, allowing for it to be applied to additional classes of problems.
Three possible methods of enhancement were identified: using the beta distribution to improve
sampling, adding functionality to solve constrained problems and, lastly, implementing
a non-dominated sorting algorithm to solve problems with more than two objectives. The
new algorithms incorporating these enhancements were developed and tested on benchmark
problems. Subsequently, the results were analysed using standard performance indicators
and compared to results produced by the original MOO CEM algorithm. The findings of this
study indicate that using the beta distribution improves sampling and therefore algorithm
efficiency. Methods of handling constraints and solving problems with an increased number
of objectives were implemented successfully. Based on these results, a final algorithm
implementing the enhancements is presented.

A Column generation approach for product targeting optimisation within the banking industry

Jean-Pierre van Niekerk — 2022-12-22

Product targeting optimisation within the financial sector is becoming increasingly complex as optimisation models are being exposed to an abundance of data-driven analytics and insights generated from a host of customer interactions, statistical and machine learning models as well as new operational, business, and channel requirements. However, given the expeditious change in the data environment, it is evident that the product targeting formulation cited throughout the literature has not yet been updated to align with the realistic modeling dynamics required by financial institutions. In this paper, an enhanced product targeting formulation is proposed that incorporates a large set of new modeling constraints and input parameters to try and maximise the economic profit generated by a financial institution. The proposed formulation ensures that the correct product is offered to the desired customers at the best time of day through their preferred communication medium. To solve the foregoing product targeting formulation, a novel column generation approach is presented that is capable of reducing problem complexity and in turn allowing for significantly larger problems to be solved to global optimality within a reasonable time frame.