These days everyone is talking about Data Science and Machine Learning – they’re the new buzzwords! How do these fresh fields compare to the seemingly age-old profession of Operations Research? What does the future hold for those in these professions? Read on to find out more about the history, current trends on these buzzwords, and what can these relatively dynamic fields learn from their older sibling – Operations Research.
Operations Research (OR) is the application of scientific & mathematical methods to the study & analysis of problems involving complex systems. It is a discipline that deals with the application of advanced analytical methods to help make better decisions, and was invented during World War II.
Data Science is an interdisciplinary field about scientific methods, processes, and systems to extract knowledge or insights from data in various forms, either structured or unstructured. It has been around since the 1960s and was known to be a contemporary data processing method.
Machine Learning is the subfield of computer science that, gives computers the ability to learn without being explicitly programmed. Machine Learning has been around as a sub-discipline of computer science since at least the 1980s.
Evolution of Operations Research, Machine Learning & Data Science over the years
Let’s look at the relative popularity of these three fields above, by leveraging Google Trends data, which is derived from Google searches that people perform; this data is a historical representation of relative volume of searches conducted on Google for the various keywords. This data helps us understand when and how the world’s interests have changed over time. The above chart compares how the relative popularity of search terms for Operations Research, Data Science, and Machine Learning has changed over time, going back to 2004 (the data is at a monthly level). In addition, the above chart shows, search interest in Operations Research has been declining steadily as compared to Data Science or Machine Learning. Note that this means that relative popularity of Operations Research is decreasing, not that the total number of searches is decreasing.
On average, people were 7 times more likely to search for Operations Research compared to Data Science in 2004. Today, they are twice as likely to search for Data Science than Operations Research. However, people were 3 times more likely to search for Operations Research compared to Machine Learning in 2004. Today, they are 3 times more likely to search for Machine Learning than Operations Research.
In 2014, there was an inflexion point where both Data Science & Machine Learning overtook Operations Research in terms of search popularity see strong evidence of both Machine Learning & Data Science picking up in popularity in recent years based on Google Trends data – let’s see if this trend holds true if we go back farther in history. For this purpose, we look at another avenue, the occurrence of these three keywords in English text books from 1950 onwards, using Google Books Ngram Viewer. The below graph shows the frequency of occurrence in published books for every year from 1950 – 2008, which is the latest year data is available.
Since Operations Research was invented during World War II, we’ve seen this field of study appear in a significant number of books, starting in the 1950’s and peaking around 1960. This strong trend continues to this day, with Operations Research appearing more than 100 times more frequently in English text books than Data Science, and almost two times more frequently than Machine Learning. Both Data Science & Machine Learning have had tremendous growth since 1998, whereas Operations Research has been flattening since that time.
We see similar trends when looking at research publications on these topics, using Google Scholar. The above graph demonstrates Operations Research trending up steadily over the years, while Machine Learning had a huge jump in research publications starting in 2000, and Data Science jumping dramatically since 2012. The number of research publications in Data Science started picking up in 2012, the same year when the number of research publications in Machine Learning started to fall.
How does the job market look like for professionals in these fields?
The above graph shows trends in job postings on Indeed.com from 2014 onwards.
Machine Learning as a key job competency surpassed Operations Research in 2014, and has been consistently higher. Data Science as a key job competency surpassed Operations Research in late 2015.
The below graph, however, shows how wages have trended since 1997 for Operations Research professionals, the only professionals whose data is tracked by the Bureau of Labor Statistics (BLS) in the US (data from BLS Occupational Employment Statistics program). We see a steady rise in both number of professionals and the annual wage, with employment numbers rising substantially starting in 2014, which is exactly when relative search popularity for Data Science & Machine Learning overtook that of Operations Research. This indicates that there is significant overlap between professionals and skillsets across Data Science, Machine Learning & Operations Research.
Indeed, the core skills remain the same, but there is just a degree of changing labels with more emphasis on predictive analytics in recent years. For example, professionals across Data Science, Machine Learning & Operations Research heavily rely on Linear Algebra for their core modeling techniques; ML models use it for solving least squares problems and matrix decompositions, Data Science models use it for fast gradient search methods in Deep Learning, and Operations Research models use it in Linear and Integer Programming. This also works from the employer’s point of view: companies looking for analytics talent require professionals who can use descriptive, predictive and prescriptive analytics to help them make better business decisions.
Whether you are a Data Science, Machine Learning, or Operations Research professional, one thing is clear – the world is full of opportunities for you!
However, there are very few companies who can consistently and sustainably recruit, retain, and grow these professionals. Having the ability to work at a company that provides challenging, yet rewarding, complex problems to solve is rare, but you can do it at Revenue Analytics.
Additionally, if you have an internal analytics team, partnering with Revenue Analytics can help you deliver game-changing analytical capabilities at lightning speed.
So, what does this mean for me?
If you are a Data Science, Machine Learning, or Operations Research professional looking for a long-term career, your best choice is joining a tech-enabled consulting firm that has one sole mission – to use advanced analytics to help businesses make better decisions, and generate organic revenue growth. Or, if you are an executive thinking about forming an internal analytics team, your best path forward is by partnering with a tech-enabled, analytically focused firm with broad experience. We at Revenue Analytics can help you apply battle-tested technology across industries to help you eliminate the unknowns and increase revenue without increasing risk.
Footnote: Data for all the charts in this blog was gathered as of June 2017