The demand for data scientists has never been greater, but how does one become a “data scientist” and what is this new career all about?
Part mathematician, part statistician, part computer scientist and part trend-spotter, data scientists are needed in a wide range of industries and fields. With vast amounts of data being collected - from finance markets and medical research to astronomy and even social media - there is a growing need for scientists to make sense of it all.
Every time we swipe a credit card, post on Facebook or visit a doctor, we create data. In addition, immense amounts of data is being gathered by climatologists, the Square Kilometre Array (SKA), geneticists, physicists and financial markets, to name but a few sources. To use this data to make useful, informed decisions requires statistical, scientific training.
UCT’s master’s programme in Data Science has been providing an interdisciplinary approach to the training of Data Scientists since 2014. Led by the Department of Statistical Sciences, and combining courses from the School of Information Technology, and several departments and the faculties of Science, Commerce and Health Sciences, the programme allows students to apply their love for quantitative and computational sciences to solving interesting real-world problems in a diverse array of applications.
“Data scientists are inquisitive: exploring, asking questions, doing ‘what if’ analysis, questioning existing assumptions and processes. Armed with data and analytical results, a top-tier data scientist will then communicate informed conclusions and recommendations across an organization’s leadership structure.” -- Anjul Bhambhri, IBM
The programme requires a good honours degree, but students who have not specialised in statistics or mathematics in their undergraduate years, can still apply for the conversion component that allows students from any discipline to pursue data science (as long as they have at least a first year statistics course and a first year computing course). The curriculum is personalised depending on the student’s background.
The number of students registered for a Masters of Science degree, specialising in Data Science (or in Advanced Analytics) has risen exponentially over the past 5 years, with many being employed by banks, accounting firms, telecommunication companies, retail companies, research institutions, and start-up data science companies. Core courses offered by the departments of Statistical Science and Computer Science teach students the art of high-performance computing, database management and data visualization, as well as the statistical methods for finding associations and patterns and building predictive models. The departments of Astronomy, Finance and Tax, Economics and Information Systems, as well as the Faculty of Health Sciences’ Computational Biology Group add the domain specializations to this skills set
The School of Informational Technology provides several courses from both the Computer Science and Information Systems departments.
Many of the core courses are scheduled for the late afternoon, catering for the needs of students who are working.
High-school learners interested in tackling data science as a career option, can register for a Bachelor of Business Science in Analytics from the Faculty of Commerce, or a Bachelor of Science in Statistics or Computer Science from the Faculty of Science.
Some examples of recent minor dissertation topics provide examples of where data science skills can be applied:
An Exploration of Media Repertoires in South Africa 2002-2014
Reinforcement Learning for Telescope Optimisation
Using a genetic algorithm approach to find optimal neural network structures for model-free policy learning in reinforcement learning
Using word2vec neural embeddings to create item similarities in grocery retailer data
Implementation of an IoT-Connected Camera for Real-Time Parking Spot Vacancy Analysis
Volume Prediction When a Mobile Data Site is Upgraded to LTE
Prediction of Mobile Network Congestion
Named Entity Recognition Using Neural Networks
Natural Language Financial Forecasting: The South African Context
Investigate a collection of machine learning models to predict and provide insights into the Net Promoter Score of a large Telecommunications company using network performance stats.
Building a conversation interface for the introduction to statistics course using Machine Comprehension of Text
Natural Language Processing on Databases
Portfolio Management with Deep Reinforcement Learning: An Application to South African JSE Top 10 Stocks