Almost all operational decision-making processes involve various uncertain parameters, and it is crucial to predict them. However, no one can really make perfect predictions, and all predictive models (parametric distributions, machine learning models, etc.) are always mis-specified – in other words, “All Models are Wrong,” a common aphorism in statistics by George Box. This is because we are only given limited information, such as a finite number of historical observations, in reality!

I study how operational decisions should be made to mitigate risk caused by such limited information on the uncertainty. In my research, some of questions I answered or am trying to address is: How should healthcare providers implement dynamic policies to protect against the risk of not satisfying service agreements under multiple sources of uncertainty? How can one make inference on uncertain parameters of service systems when available data only reveals indirect information? If a retailer is competing with others without (fully) understanding demand uncertainty, how many orders should be placed? Just like these, there are many, many questions to be answered, and this is why I love my research. 😊

I am always interested in collaborations! Please reach out to me to share your thoughts and engage in inspiring discussions 😃