Speakers Bio & Abstract

Biography
Peter Leihn the Director of Business Development and Commercialisation at Data61 (formerly NICTA), a CSIRO company. Peter has worked across Australia and Asia with global ICT companies Hewlett-Packard and Autodesk developing markets for new technologies. Peter held leadership positions with NSW Government including heading the Office of the Director-General for Water and Energy, Director, Office of the Chief Scientist and Engineer and Director, Office for Science and Research.Abstract
Spatial Predictive Analytics: Extracting Insight and Maximising ImpactFederating and visualizing spatial datasets is only the first step toward unlocking the knowledge they contain, as demonstrated by Terria Maps, an interactive, open sourced, 3D data visualisation tool. Building on the success of this software, the team is further developing a suite of Analytics tools to be utilized in a Terria Map Deployment. These tools are built from state of the art machine learning algorithms designed specifically for large scale spatial inference. The breadth of Terria?s spatial analytics capacity is demonstrated through the following capabilities: ? Spatial detailing - often spatial datasets are released as averages over regions that may be too large to provide any useful information. By compiling related datasets with higher detail level, Terria can automatically infer their statistical relationship and estimate the original dataset in greater detail. ? Spatial Community Discovery ? Terria?s machine learning algorithms provide an automated approach to summarise complex datasets objectively, allowing maximal information to be extracted ? Spatial Prediction and Active Sampling - Using inexpensive or easily acquired data as a proxy is an efficient way to build models. Terria?s algorithms take a probabilistic approach by providing both a predictive model that fuses multiple data sources and a measure of how certain this model is at each point in space. This allows the user to quantify the risk of a decision and to automatically determine where and how to sample in the future. ? Spatial Demand Distribution ? Terria?s specialized spatial regression algorithm is able to predict variations in demand for products and services by relating sparse sales data with a variety of spatial demographic information. The above sample of tools created within the Terria team can be applied to a wide range of fields and assist in extracting value from datasets that would otherwise be overlooked due to their complexity and size. The unification of these analytics tools with an interactive map deployment becomes a powerful, convenient and highly impactful tool for businesses, research and the like.