Feature Learning for High-dimensional Functional Time Series

The proposals outcomes will enhance modern statistical analysis and benefit empirical researchers given the easy-to-implement algorithms the research will advance. This project drives broad benefits across Australias priority areas because it has diverse applications on multi-sectional complex data. For instance, optimal stopping time for multi-section wood panel compression in advanced manufacturing, accurate forecasting for multi-country mortality forecasting in demography, large financial portfolios in finance, multi-site temperature forecasting in environment science, large-household daily electricity consumption in economics. This promotes strong statistical techniques in various sectoral areas and leads to more accurate forecasting outcomes and novel data interpretations for defining features. Research workshops and open access dissemination will connect project outcomes to both researchers and practitioners.