Published on 15 Jul 2019

​Chat with the Professor: Neumann Chew – Senior Lecturer, Nanyang Business School

Neumann Chew shares more of his work in NBS and views on data analytics.

Currently, Neumann is teaching Business Analytics and finishing his textbook to consolidate more than 15 years of teaching and best-practice first-hand Analytics problem-solution consulting for his students. We have the pleasure of having Neumann join us at the TECHINSG V2.0 event on 18 July 2019 sharing his views and hopefully work on AI and RPA. Before that, we spoke to him, and found out more about him, his work and views.

TFAGeeks: “Hi Neumann, thank you for speaking to us. We know that you are currently a senior lecturer at Nanyang Business School and is teaching Business Analytics to aspiring undergraduates. You are the author of the 3 volume textbooks – Analytics, Data Science and Artificial Intelligence. Previously, you were the Principal Consultant (Analytics) at SAS institute, where you worked with well-known clients to design and deploy Analytics solutions. Can you share with us more about your passion in statistics and analytics, what is so intriguing about the subject?”

Neumann Chew: “Solving Real World Problems and Opportunities using Data. Most individuals and organizations do not lack data. We need more insights from the data. The fusion of Statistics, Analytics, Machine Learning and A.I. enable new ways of analyzing data and generating additional insights.  The more data we accumulate, the greater the need for insights and data-driven decision making, regardless of industry. We are proving the value and benefits of Analytics, Machine Learning and AI to more and more clients, and inventing new techniques to analyze data. This is exciting work.”


TFAGeeks: “We are aware that Data Science, Data Analytics and Artificial Intelligence can solve many issues and problems for society, corporations and government. In your opinion, what is the most underrated field or focus area that many have missed out?”

Neumann Chew: “Transparent models. The common way to consider effects of multiple variables simultaneously is to use a model and thus, there are many models in Statistics, Data Science, Analytics and A.I. In recent years, Deep Learning became popular, advanced by Google. However, Deep Learning and in general, Neural Networks, are black box models. The link from inputs to outputs are not direct and obvious to non-technical users. This is dangerous. What if the decisions, based on the outputs of a black box model, lead to big losses or death to the patient?”“There are more transparent models such as Classification and Regression Trees, Random Forest etc, whose workings are more direct and obvious, but less “famous” and less well known in industry compared to Deep Learning. This is one reason why I am publishing my textbooks – to explain and show the merits of transparent models such as Classification and Regression Tree, Random Forests, MARS (Multivariate Adaptive Regression Splines) etc.”

TFAGeeks: “Do you think the current trend and fiery interests in Data Analytics and Artificial Intelligence is a fad and soon it will popped like a dot com bubble?”

Neumann Chew: “I see the developments as an evolution. When I was an undergraduate 20 years ago, the more common term was Data Mining. As individuals and organizations grow in their collective intelligence capability and more importantly, trust in the transformative power of Analytics and/or AI, they will want to do more with their own Data. This has already happened in many organizations globally, and in Singapore – DBS, IRAS and others. The interests will only rise with each successful application, though we may have a new term 5 years later, perhaps Generative Intelligence.”

TFAGeeks: “We heard about your work in automated compliance checks and red flagging with Machine Learning, and are intrigued to know more about it. Can you share more about it, how it works, and how it can save time and efforts for organizations which have adopted it?”

Neumann Chew: “The stimulus came from consulting work with CEOs, managers and auditors who are searching for more effective and efficient way to check for non-compliance to policy and procedures, and suspicious cases (e.g. fraud). The standard sampling method does not consider all cases, take too long and too dependent on human judgement. I adapted my previous research in Anomaly Detection and developed a Machine Learning methodology that scans all cases and automatically flag out non-compliance and suspicious cases. This is far more comprehensive, faster and effective than sampling method and can be used to identify high risk areas before it is too late. It can be used as a continuous, tireless risk scanning tool to check all records and prioritize cases for further investigation.”

TFAGeeks: “If there is one particular Analytics project that you can create and implement with all the resources granted, what will it be and why?”

Neumann Chew: “Integrated A.I. for a mission critical application in an organization that truly believe and embrace the transformative power of A.I. To my knowledge, no companies had achieved this pinnacle. A.I. is not new and some advanced companies had A.I. capabilities but these are often siloed applications and not fully integrated and self-managing in operations, processes or strategy decision making. At the pinnacle, A.I. will take over the entire process and will self-learn, self-adjust, self-correct, self-optimize from start to end, without human intervention. Humans can provide feedback to A.I. but no longer need to watch and manually intervene. Eventually, human feedback will be unnecessary. The most technologically advanced companies are exploring and learning how to achieve this. Developing advanced predictive model is necessary but insufficient. Currently, such models still require human experts to check and improve.”

“As of now, the closest application to this pinnacle is Self-Driving car on any road without any human intervention. This technology is still in development and a race by autonomous vehicle companies. There are limited success in controlled areas e.g. industry parks, human driver-led chain of self-driving long distance trucks, etc. In more general roads, deaths were reported in some trials.”

“I am interested in developing such an integrated system regardless of industry or application. The most difficult obstacle is actually human. Many people are more comfortable blaming PDPA, Privacy Laws, Company Policy, etc. All these are safeguards that can be accommodated in Analytics or A.I. projects. Using these as excuses to stop or delay just shows the lack of human will and belief in the power of Analytics or A.I. I had seen so many that have the potential to re-write history with Analytics or A.I. in their organization, but do not have the courage or will to do it. We only live once.”

“Below is my chart that shows the growth in collective intelligence and the competitive advantage it endows.”


TFAGeeks: “Last question, do you sleep at night? Or you dream about coding and refining your works?”

Neumann Chew: “I enjoyed sleeping. 10 hours is optimal for me. In fact, I solved some of the most complex problems during sleep. In several occasions, I jumped out of bed and wrote down the solution to the problem yet to be solved the night before. To me, coding is just an execution of a precise plan. The plan with all the ideas and concepts to solve the problem is far more important. The computer and code is just a tool to express and/or test our ideas. I don’t dream about coding. I dream about the problem/opportunity and solution, sometimes.”

Source: TFA Geeks, 15 July 2019