Researcher:  Shirindi Ntshuxeko, Sol Plaatje University
Supervisor: Dr Colin Chibaya, Sol Plaatje University

A formal knowledge domain has not been well represented in earlier studies.  There has been a lack of a particular set of procedures required to produce an ant ontology.  To create an ant colony ontology, this work aims to discover and define the fundamental traits of a simulated ant system.

Researcher:  Eli Nimy, Sol Plaatje University
Supervisor: Dr Moeketsi Mosia, Sol Plaatje University

In recent times, universities have become greatly dependent on the use of online educational platforms (OEP) such as Moodle, Blackboard Learn and Canvas to share resources, assess, and communicate with students. But how can we support students that are not engaging and engaging with these platforms differently?

Researcher:  Mbabala Tshimangadzo, University of Venda
Supervisors: Dr. N.E Maluta , Prof. R.R Maphanga Mr. R.S Dima, University of Venda

There is an increase shortage of energy supply and storage, with the human population and fuel price increasing exponentially this increases the demand of energy supply. Considering battery as the source of energy supply will require the discovery of materials but the traditional method of doing this is not only expensive but also very slow. This project applies machine learning algorithm which will accelerate the discovery of battery materials with the desired properties which is quick and cheaper, this will not only helps us in our homes but also in our vehicles.

Researcher:  Wandile Nhlapho, University of Venda
Supervisor: Dr Jean-Claude Ndogmo, University of Venda

The study compares the performance of the ARCH (1) and GARCH (1,1) models in estimating and forecasting the volatility of Telkom share prices.  The Telkom shares are estimated using daily data and the above-mentioned volatility models. We estimate our models using the normal (Gaussian), student t, and generalized error (GED) distributions to determine which distribution best fits our models. Log-likelihood, Schwarz information criterion, Hannan-Quinn information criterion, and Akaike information criterion were utilized to evaluate those distributions. We forecasted our models using the distribution with the lowest Akaike, Schwarz, Hannan-Quinn, and log-likelihood values. Theil’s inequality coefficient, mean absolute error, and root squared mean error are three forecasting evaluation measures used to assess the model’s forecasting performance.

Researcher: Mawela Vhutshilo, Sol Plaatje University
Supervisor: Dr. Colin Chibaya, Sol Plaatje University

It has taken billions of years for nature to evolve. We can create systems that are considerably more effective and efficient by mimicking nature. When it comes to product manufacturing, we are dealing with a wide range of problems. The manufacture of a few products due to a lack of necessary machinery is one major problem. By studying the behaviors of birds, such as how birds disperse and congregate together to form a coherent behaviour. We can create software that copies these behaviors by using several robotic machines in place of people to assist in the simultaneous production of various products.

Researcher: Costa Muthai, University of Venda
Supervisor: Dr Martins Aramsowna, University of Venda

Internet and Web technologies of today not only enable students to interact more freely with educational resources, friends, and teachers, but they also produce enormous amounts of application data that can be assessed to reveal study and learning habits. The Kalboard 360 Learning Management System (LMS) data was used in this research study to analyze student trajectory data from a blended learning course and create a probabilistic (Bayesian) model that predicted academic success. Statistical inferences were made to classify students and highlight characteristics from the data that corresponded to failure based on the influence of their demographic, academic, and behavioral characteristics.

Researcher: Mukondeleli Nengwani, Sol Plaatje University
Supervisor: Dr Colin Chibaya, Sol Plaatje University

Animals have a way of working together cooperatively that helps them achieve certain goals in fish swarm, the fish have individual tasks that assist in the success of their schooling.  An investigation of those tasks is of interest in this study and the behaviors were found and interpreted computationally.

Researcher: Kelly Langa, University of the Witwatersrand, Johannesburg
Supervisor: Prof. G Nwaila, University of the Witwatersrand, Johannesburg

Uncertainty Quantification in Global PGM Production using Stochastic and Machine Learning Forecasting Algorithms.  The applications of platinum group metals (PGMs) are innumerable and stretch across multiple industries due to their mechanical and chemical catalytic properties.  The introduction of Industry 4.0 and the many challenges associated with mining along with changing world economic systems is causing massive changes in the supply chain of PGMs.  This has consequently led to erratic supply patterns and deficit of PGMs. The growing uncertainty in global PGM production raises a need for the development of robust and efficient data driven PGM production forecasting methods. Accurate production forecasting is indispensable for mitigating potential supply chain disruptions and strategic planning. The advances in technology has brought about new advanced data analysis techniques which presents us with the opportunity to apply them to the most sophisticated of tasks.

Researcher:  Stuart Morrison, University of the Witwatersrand, Johannesburg
Supervisor: Prof, Rod Alence, University of the Witwatersrand, Johannesburg

In this work I seek to understand the effect that fatalities have on the accuracy of election violence datasets, by asking “what happens to the quality of the dataset if we create a dataset that includes non-fatal incidents?” The results from the study show that including non-fatal events can decrease the accuracy and overall quality of the dataset by including too much noise and that the fatalities themselves help in describing how the actors interact with each other.

Researcher: Sizo Mosibi, University of the Witwatersrand, Johannesburg
Supervisor: Prof. Yudhvir Seetharam, University of the Witwatersrand, Johannesburg

Predicting future asset prices with higher accuracy has remained a formidable undertaking. Series of techniques have been developed and implemented to attain the intended result, and significant progress has been made. The use of a sentiment index as an independent variable in a machine learning model to predict asset prices, which utilizes long short-term memory, is one of the techniques employed. We developed a sentiment index to augment the South African economy and included it into the LSTM model to estimate future Sasol prices. We discovered using a sliding window strategy that using raw price in the LSTM model beats the inclusion of investor sentiment index as a variable. We also observed that the arbitrage pricing model (APT) performed the worst at predicting asset returns because it failed to account for non-linearity in price evolution.