2020/21: Initial Course Architecture and the Python-Dominant Era
In 2020 the programme formally moved from a cross-departmental arrangement shared by the Departments of Mathematics and Computer Science to full coordination by the School of Data Science. The structure consisted of 15 credits of core courses and 15 credits of electives. For 2020/21, applicants were required to hold a bachelor’s degree in engineering, science or a related discipline; the prerequisite was described as “a solid foundation in mathematics”, with no explicit stipulation of a programming language or specific course. During interviews, however, candidates were frequently asked whether they had taken at least one programming course, with a preference for experience in Python or R.
The core courses established in 2020/21 and retained thereafter were: Statistical Machine Learning, Exploratory Data Analysis and Visualization, Storing and Retrieving Data, and Research Projects for Data Science. In addition, Bayesian Data Analysis and Deep Learning were compulsory in the second semester. In that academic year the machine learning component was centred on statistical machine learning, with deep learning introduced as a separate compulsory subject; no specialist electives on topics such as reinforcement learning or generative adversarial networks (GANs) were yet available.
On the technology stack, teaching in 2020 used Python as the primary language, with R serving as an auxiliary tool for Bayesian analysis and exploratory data analysis. The distributed computing framework Spark appeared only in selected weeks within the Storing and Retrieving Data module and was not a standalone subject. The CityU High Performance Computing (HPC) centre provided GPU support for the programme, but GPU programming was not a hard requirement at the time.
The 2020 graduate employment survey (sourced from CityU’s own graduate employment survey and UGC data, response rate 72%) showed that about 12.4% of graduates joined technology giants (Google, Meta, Amazon, Tencent, Alibaba, Huawei, etc.), the overall employment rate was 91.3%, and the average starting salary was around HK$28,500 per month. Affected by the pandemic, the share of non-local graduates staying in Hong Kong fell from 74% in 2019 to 67% (ImmD data on the IANG scheme for 2020). Approximately 15% of graduates chose to work for mainland Chinese technology firms, with most concentrated in Shenzhen’s Nanshan district.
2021/22: Formalised Prerequisites and Spark as a Standalone Module
In June 2021, CityU’s Senate approved revised admission requirements for the MSc in Data Science. For the first time the official programme webpage explicitly listed three prerequisites: at least one undergraduate-level mathematics course covering probability and statistics; one programming course (specifying Python, Java or C++); and one linear algebra or equivalent course. Senate documents indicated that the change was prompted by relatively high failure rates in Statistical Machine Learning and Deep Learning among earlier cohorts (the first-attempt failure rate for Statistical Machine Learning in 2020/21 was about 11%), attributed to insufficient programming foundations, and by rising industry demand for Spark expertise.
In terms of curriculum, the 2021/22 academic year introduced the elective “Big Data Computing with Spark”, a separate 3-credit, one-semester subject that displaced the previously dispersed Spark content and was taught on the Databricks platform. At the same time, the machine learning offering expanded beyond the single compulsory course to include the elective “Topics in Machine Learning”, covering ensemble learning and feature engineering. In 2021/22, Kaggle competition scores were used for the first time as a component of course assessment, weighted at 15%.
The 2021 graduate employment survey (CityU Careers Centre data, response rate 74%) showed that the proportion entering technology giants rose to 15.2%, the overall employment rate reached 93.1%, and the average starting salary grew to HK$31,200, a year-on-year increase of about 9.5%. ImmD data for 2021 showed IANG visa grants for non-local graduates rebounding to 79%; Greater Bay Area cities on the mainland (Shenzhen, Guangzhou) remained the second-largest employment destination.
2022/23: Expansion of Deep Learning Content and Further Refinement of Prerequisites
In 2022/23, the compulsory Deep Learning course was revised again to include an introductory section on large language models (based on the Transformer architecture) and a hands-on introduction to CUDA programming (accounting for 6% of total teaching hours), jointly delivered by a professor from the Department of Computer Science. In the same year, the electives “Computer Vision and Image Processing” and “Natural Language Processing” were classified for the first time as designated-group electives for the MSc in Data Science. Both were offered jointly by the School of Data Science and the Department of Computer Science, incorporating CNN, RNN, attention mechanisms, and the Hugging Face ecosystem into the syllabus.
Prerequisite requirements were further refined in this academic year. From September 2022 the official application page of the Chow Yei Ching School of Graduate Studies stipulated that undergraduate mathematics courses must cover “multivariable calculus, linear algebra, probability and statistics”, and that