Predicting Particle Fineness in a Cement Mill

Researcher: Rowan Lange, University of the Witwatersrand, Johannesburg
Supervisors: Prof. Anton van Wyk, Dr. Terence van Zyl

Cement production is a multi-billion dollar industry, of which one of the main subprocesses, cement milling, is complex and non-linear. There is a need to model the fineness of particles exiting the milling circuit in order to better control the cement plant. This paper explores the relationship between the particle size of cement produced and various sensor readings from the cement mill circuit. The aim of this paper is to provide a model for predicting the fineness of particles exiting the milling circuit using data on the current and past states of the plant. A comprehensive literature review of the problem as well as a discussion of potential modelling solutions is provided. Blaine (particle fineness) is modelled using many different linear and non linear models on 5 months of data from a large cement plant. On a holdout test set a multi layered perceptron achieved a MAE of 8.799 and a linear regression achieved a R2 of 0.481. discussion of the significance of various features for predicting Blaine is also presented. The results show some success from non-linear data-driven models and highlight the unique difficulties in modelling the cement mill, presenting recommendations for future research.

View the full report

Read More

Indigenous Women Moving From Physical to Digital Fires: The Evolution of Methods of Transmission of Indigenous Knowledge

Researcher: Khanyisile Yolanda Ntsenge, University of the Witwatersrand, Johannesburg
Supervisors: Dr Constance Khupe and Prof. Rod Alence, University of the Witwatersrand, Johannesburg

In response to the threat of extinction of indigenous knowledge, there has been a growing number of people, a significant amount of whom are women, interested in the preservation of indigenous knowledge systems who have begun to use social media platforms such as Twitter and YouTube and the indigenous method of storytelling to share indigenous knowledge.
The aim of the study is to understand how the introduction of the social media platforms Twitter and YouTube has changed the community structures for sharing indigenous knowledge in physical versus social media communities.

The research is informed by a postcolonial indigenous and indigenous feminist approach and employs transformative participatory research in its methodology. Indigenous women in physical communities participated in the research while accounts owned by indigenous women on Twitter and YouTube were analysed. A social network analysis was conducted on both the physical communities data and social media data. Sentiment analysis was conducted on the social media data.

The results show that the network of communities while both anchored by indigenous women have different structures. The physical communities were very tight-knit with members of the networks learning and sharing indigenous knowledge amongst each other thereby potentially reinforcing their knowledge. The social media communities were mainly connected only to the main account and members rarely engaged with each other. The sentiment analysis found conversations in the social media networks to be significantly positive with the highest scoring emotion being that of trust.

The research has shown that although women play an important role in the sharing of indigenous knowledge in both physical and online communities, the community network structures differ. It also evidenced that there is a space and appetite for conversations on indigenous knowledge on social media. Furthermore, as they are in physical communities, women continue to be important custodians of indigenous knowledge and are trusted to share credible indigenous knowledge. This presents opportunities for further exploration on how to leverage social media platforms to mainstream indigenous knowledge while amplifying the voices of indigenous women as custodians of indigenous knowledge.

View the full report

Read More

Using discrete emotions and review discrepency to predict the helpfulness of mobile app reviews

Researcher: Mpho Modise, University of the Witwatersrand, Johannesburg
Supervisor: Dr S. Verkijika, University of the Witwatersrand, Johannesburg

In our digital lives, smartphones have become indispensable tools, and the Google
Play Store stands as a central platform for mobile applications. User reviews on this
platform significantly influence consumer decisions. Nevertheless, the sheer volume
of reviews poses a formidable challenge. To address this issue, this study suggests a
comprehensive approach employing algorithms to assess the quality, relevance, and
credibility of reviews.

 

Read More

Safe urban futures: exploring the nexus between urban safety and quality of life in South African cities

Researcher: Lerato Tlale, University of the Witwatersrand, Johannesburg
Supervisor: Helen Sarah Robertson, University of the Witwatersrand, Johannesburg

Urban safety’s correlation with holistic quality of life in South Africa’s evolving urban landscape demands comprehensive exploration and understanding.
South Africa’s population, estimated at 58 million in2019 and projected to reach 65 million by 2030 and 70 million by 2043, underscores the significance of quality of life amid rising urbanization rates.
The exploration of the relationship between elements of the built environment, like urban safety, and quality of life is pivotal in shaping present and future urban developments in South Africa.

 

Read More

Temporal Dependency Modeling in Financial Markets

Researcher: Small Tshithavhana, University of the Witwatersrand, Johannesburg
Supervisor: Dr. Walter Mudzimbabwe, University of the Witwatersrand, Johannesburg

Financial forecasting has become increasingly important in today’s global market due to its ability to
assess risk and inform decision-making. However, accurately forecasting financial markets is challenging
due to their stochastic nature and complexity. To address this challenge, we suggest a state space
model, namely the Hidden Markov Model, which handles dynamic time series issues involving unseen
variables or parameters that represent the development of the underlying system’s state.we test
our model on financial market information sourced from the Nasdaq online database and compare its
performance with standard forecasting machine learning models. The results under the MAPE matrix
indicate that the proposed model outperformed the Recurrent Neural Network (RNN) by 19.08% and
exhibited a superior performance of 19.09% relative to the ARIMA model. However, the proposed
model fell short in comparison to the GARCH model by a margin of 3.11%.

 

Read More

Comparison of ensemble nlearning and neural network models for default risk analysis: a reproduction study

Researcher: Tebogo Malatsi, University of the Witwatersrand, Johannesburg
Supervisor: Assoc. Prof. YudhvirSeetharam, University of the Witwatersrand, Johannesburg

This study replicates prior research on default risk assessment using ensemble and deep learning techniques, leveraging payment data from the UCI Machine Learning Repository for Taiwan. It specifically compares boosting, random forest, and neural network models with Tanh and ReLUactivations, evaluating their predictive accuracy and classification capabilities through metrics like accuracy, AUC, ROC curve, and F-Score.
The study reveals trade-offs in model performance metrics, with heightened accuracy but nuanced reductions in AUC/F-Score. Random Forests demonstrate superior learning and generalization, while neural network models emphasize the challenge of balancing sensitivity, specificity, precision, and recall.

 

Read More

Email monitoring system using machine learning

Researcher: Netshamutshedzi Ndivhuwo, University of Venda
Supervisor: Dr Ibidun Obagbuwa, University of the Witwatersrand, Johannesburg

This study investigates email monitoring systems using machine learning. In this study, we contribute to previous studies on spam problems to improve accuracy by utilising a variety of methods.

 

Read More

Ever wondered what to use to make the right investment decision?

Researcher: Dimpho Mosaka, University of the Witwatersrand, Johannesburg
Supervisor: Dr Martins Arasomwan, University of the Witwatersrand, Johannesburg

More studies have demonstrated that the DeepNeural Networks(DNN)techniques can be used to forecast the stock market price. This research project aims to investigate the effectiveness of deep neural network techniques in the SouthAfrican stock market price forecastinhusing Long-Short TermMemory(LSTM) and Convolutional Neural Network( CNN) technique, including the challenges and opportunities of this approach

 

Read More