IFin Seminar, Jin-Chuan Duan, National Univesity of Singapore - "A New Data Analytics on Small or Large High-Dimensional Observations in Finance and Beyond"
Istituto di finanza
Data: / -
Tema del Seminario: "A New Data Analytics on Small or Large High-Dimensional Observations in Finance and Beyond"
Prof Jin-Chuan Duan, National Univesity of Singapore
12:25 - 13:40
Auditorium (stabile principale, 3° livello)
Campus di Lugano
Data analytics (DA) in the modern context is often understood as the processes and tools for extracting information out of Big Data. A tough challenge in DA is to deal with high dimensionality and/or non-linearity of the data rather than large size as measured by the number of data instances. Considering interaction terms is, for example, natural in many research and application contexts, but that makes dimension of the data much larger, and in essence also turns a model non-linear, without a corresponding increase in the number of data points. Neural Networks and many other machine learning tools are often ill-suited for DA problems of this type that are not always accompanied by a very large number of data instances. I will introduce a new variable selection tool based on a zero-norm penalty which can also be understood as a kind of interpretable AI. I will demonstrate with examples how this new technique can handle high-dimensional data in finance and beyond where the number of observations is either large or small. The talk will also touch upon the AI work-in-progress assisted with this new analytics at the Credit Research Initiative of the National University of Singapore.