Researcher: Tshilisanani Mudau, University Venda
Supervisor: Dr Michael Klipin and Prof. Hima Vadapalli, University of the Witwatersrand, Johannesburg

Machine Learning models have become the modern way to protect patient health information. Which Machine Learning model can best perform this task?

The Conditional Random Fields model easily recognized sensitive Health Information when given more or less information. Random Forest only recognized sensitive Health Information when more information is given. The patient sensitive information learnt by models is replaced with fake but meaningful information to protect the owner of the information from being traced.

Researcher: Gift Mahlatse Mphahlele, University of the Witwatersrand, Johannesburg
Supervisor: TBA

This study aimed to investigate the nature of the relationship between the global news sentiment with the local news sentiment and exchange rates for two countries, South Africa and Nigeria.  The transformer architecture was used to do sentiment analysis through a sentiment analysis pipeline.  The Pearson’s correlation test showed that there is no significant relationship between the global news sentiment with the local news sentiment and exchange rates for both the countries News Bias and Fairness in Emerging Markets.

Researcher: Christina Meletakos, University of the Witwatersrand, Johannesburg
Supervisors: Prof. Rod Alence and Dr Ekeminiabasi Eyita-Okon, University of the Witwatersrand, Johannesburg

Questions were asked as to the role corruption plays in the building and production of the following renewable energies (RE) across the world: Hydropower, Biopower, Wind power, & Solar energy. Multiple regression was used to understand this relationship.

Researcher: Nkosenhle Mdluli, University of the Witwatersrand, Johannesburg
Supervisors: Rudzani Mulaudzi and Dr Ritesh Ajoodha, University of the Witwatersrand, Johannesburg

As of quarter 2 2021, South Africa officially became the country with the highest unemployment rate in the world. The rate currently sits at 34, 4% and is the highest it has ever been in the recorded history of the country. To better inform policy decisions that can reduce this rate, it is important that the structure of the South African labor market be understood.

Bayesian networks, a type of probabilistic graphical model, were used to construct a structure showing how the unemployment rate is related to other macro-economic variables. Three models were constructed using the hill climbing algorithm with the best scoring model attaining an auc score of 0.895. These models were used to perform inferences on the unemployment rate. An increase in FGE was found to have the most desirable impact as it decreased the UR the most.

This research thus provided valuable insight into the labor market structures and will help act as a reference point for researchers and policy makers in South Africa.

Researcher: Neo Matsobane, University of the Witwatersrand, Johannesburg
Supervisor: Dr Wilbert Chagwiza, University of the Witwatersrand, Johannesburg

The idea of predicting stock prices has always appealed to both financial investors and researchers. The stock market is unforeseeable in nature whereby financial Investors consistently enquire if the cost of a stock will increase or not. New technologies like data mining, machine learning and deep learning helps to examine large information and build up a model that keeps away from human mistakes during stock predictions. The purpose of this study is to build a recurrent neural network (RNN), specifically long-short-term memory model (LSTM) that predict stock market.

Researcher: Rosa Manoim, University of the Witwatersrand, Johannesburg
Supervisor: Dr Michelle Small, University of the Witwatersrand, Johannesburg

According to the South African government, “the advancement of human rights and the promotion of democracy are pillars on which South Africa’s foreign policy rests”. Yet as the statistical results here show, this commitment towards respecting and promoting human rights is not reflected in South Africa’s arms exports.

Researcher: Rinae Tshivhidzo, University of the Witwatersrand, Johannesburg
Supervisor: Dr Wilbert Chagwiza, University of the Witwatersrand, Johannesburg

Researchers have devoted their focus to studying stock market prediction. However, predicting the stock trend is still an open question. With the development of the advanced period, the prediction of stock market has moved to the field of machine learning. In many kinds of research that have been conducted, machine learning techniques have proven to be robust in predicting stock market. This research aimed to train a model that accurately predicts stock market based on fundamental analysis using a random forest algorithm. The model is trained on the US stock market using fundamental analysis. The results show that the random forest model succeeds in the prediction of stock market. Hence random forest was able to predict stock market with a 0.93 degree of accuracy.

Researcher: By Kabo Phage, University of the Witwatersrand, Johannesburg
Supervisor: Prof. Gregory Farrell, University of the Witwatersrand, Johannesburg

Financial cycles capture the evolution of risks to financial stability, and it follows that they are important for macro prudential policymakers.  The robust measurement thereof can aid in formulating and implementing policy. This report adds some new evidence to scarce South African literature by focusing on measuring the financial cycle using Continuous Wavelet Transform techniques, which can decompose a time series into statistically significant frequency ranges used to identify cyclical behaviour. The results show that the South African financial cycle is well defined when identified by the co movements of medium-term cycles in credit and house prices whereas equity prices tend to be less informative. Furthermore, the financial cycle is longer in duration than the traditional business cycle and so policymakers should focus monitoring on the medium term for the sake of identifying the buildup of risk.

Researcher: Sheena Phillip, University of the Witwatersrand, Johannesburg
Supervisor: Dr Wilbert Chagwiza, University of the Witwatersrand, Johannesburg

Fraud detection of health care providers is a growing concern worldwide as billions of dollars is lost each year. Medicare publicly released health provider data in order to encourage the development of models to overcome fraud. The aim of this research is to train a model using the Medicare dataset to determine with what accuracy the model can predict fraud and to identify the top 5 features which contribute the most towards fraud detection. It also aims to investigate the impact that explicit features such as Provider, BeneID and ClaimID have on the accuracy of the model. Four datasets were combined into a single comprehensive dataset and was subsequently used to train an XGBoost model. The model had accuracy of 0 98 with a recall score of 0 97 and performed extremely well overall. The model trained on the dataset excluding explicit features produced an accuracy of 0 85 and a recall score of 0 71. Comparatively, the model performed poorly with a 13 drop in accuracy. It is noted that regardless of which feature space was used, the top 5 features encompassed details about the doctor as well as the location of the hospital.

Researcher: Ruan Pretorius, University of the Witwatersrand, Johannesburg
Supervisors: Prof. Terence van Zyl, University of Johannesburg and Dr Farai Mlambo, University of the Witwatersrand, Johannesburg

Accurate financial time series forecasts can assist investors in gaining a competitive edge over other participants in capital markets No empirical conclusion existed on what the most accurate model(s) were for forecasting stock market price movements over different forecast horizons Limitations from previous studies were addressed in this study by compared the forecasting accuracy of 20 different models on 403 time series of stocks/indices These included machine learning ( statistical and benchmark models The naïve benchmark model outperformed all other models in this study for nearly all accuracy metrics and forecast horizons tested.