Operational research, statistics and big data

The Department of Management Science at City University is composed of operational research and statistics. However, many years ago, when I studied for a Masters of Operational Research at the University of Lancaster in the United Kingdom, statistics was considered part of operational research. it is impossible to apply the mathematical models of operational research without the basis in statistics.

In the earliest days, operational research was the application of mathematics to operation management, which was a branch of mathematics. It was only when I entered the mathematics department in the 1970s for my undergraduate degree that I came across the term operational research. In the early days, when I was teaching at City University, many students who like me, first learnt about operational research when they entered university, asked the question "What kind of job can I find after graduating from operational research?"

Back then, when operational research was still a branch of mathematics, the actual industrial or commercial use of operational research was indeed not extensive – as the ability to calculate was rudimental. Many operational research model calculations were impossible to calculate by hand or by a simple calculator. Even further back in the 1960s when I studied middle school, the calculator which started to be available to the public was very expensive; most calculations back then were still conducted with a pen and paper. In other words, the application of operational research needed to be matched by a similar ability to calculate, which is why the popularisation of computers was so important for this branch.

By the early 1980s when I returned from England to Singapore to work in the F&B and Agro-product sector, my first challenge was to develop a computer calculated formula for pig feed. This was a classical applicational of linear programming which was taught during university. However, back then, in exam questions there wouldn't be more than four variables as anything more complex could not be easily calculated by hand or by a simple calculator.

Coincidentally, in the 1980s, Apple launched its Apple II computer and the company decided to purchase one, a computer with only 64K of computing power. That is 64k, not 64M, or 64G! Back then I could not find any linear programming software, so I wrote myself one with BASIC programming language. Guess how long it took to calculate at that time? What can be calculated in less than one second today took one hour back then. In the early 1980s, the writing of my own software and using Apple II to calculate feed formulas for pig farms was considered a big deal in the eyes of my boss. However, my identity in operational research and computer science was inseparable back then.

Many years later the term big data emerged. In the stock market, the hype was extremely high. These stocks were called new economy stocks. When e-commerce platform Meituan was preparing to go public in Hong Kong and I was asked to sit in with management to understand more about the company, the management talked about big data and the best route for management to achieve the most effective meal delivery service. Isn't the best route management also the most classic application of operational research? Of course, a lot of data analysis and statistics are required for application, hence the term big data overshadowed operational research, however, operational research and statistics remain the backbone of big data.

It is indeed difficult to draw a clear boundary between different schools of learning.