FREE EVENT: Next week the Commonwealth’s Data Analytics Centre of Excellence Community of Practice will be hosting a half-day teleconference on data tips and tricks for all tiers of the public sector.
The DACoE is a forum within government that develops policies and standards, shares ideas, exchanges information and provide guidance, support by the Australian Taxation Office.
“Data without ingenuity is like a lamp without power,” said Prime Minister Malcolm Turnbull at the recent GovHack Event.
When you combine the PM’s statement above with the September 2 release by the Department of the Prime Minister and Cabinet and the Australian Public Service Commission of the “Data skills and capability in the Australian Public Service”, the DACoE thought this is the perfect opportunity to fulfil its mandate to support, inform, and enhance the skills of the Australian public service, at all tiers of government.
With this in mind, the DACoE is conducting a mini-conference on Monday November 14. Attendance is by teleconference, a physical meeting venue at the ATO offices in Sydney (52 Goulburn St), and a possible video broadcast on a social media channel.
The intent of the mini-conference is to explore:
- Digital profiling
- Natural language processing
- Cohort analysis
- Digital fingerprinting
- Context analysis
- Different analytical techniques
- Building successful data analytics capability
9.00 AEDT — Opening Richard Collis (Assistant Commissioner) Smarter Data, ATO
9.10 AEDT — Presentation by Warwick Graco — Digital fingerprinting
This presentation will cover the use of configuration mining to discover the signatures of cases of interest in a population such as those who gamble irresponsibly, those who are dangerous drivers and those who light fires that cause damage to life and property. Each signature can be represented as a configuration of scores. Configuration mining can be employed to identify these signatures in data. In addition affinity mining can be applied to data to discover indicators that are proxy measures of particular traits and tendencies. For example, people’s use of telephones can indicate whether they will pay their debts. Similarly people’s TV viewing habits can reveal which political party they will vote for, what health issues they will have and whether they will be safe drivers. These proxy indicators can be combined with the signatures to provide profiles. Typologies, which represent recurring patterns in the profiles, can be identified such as the different types of arsonists in the community. The profiles of citizens can be compared to the typologies to see where there are matches. These matches can assist governments to identify threats such as those who will drink heavily and drive their vehicles so that preventive action can be taken thus reducing the economic and social costs to society. The ethical, privacy and security implications of digital fingerprinting will be briefly discussed.
9.30 AEDT — Presentation by Tony Nolan — Using cohort analysis with Open Source Datasets for strategic, operational, and tactical purposes.
Cohort analysis is a digital profiling technique developed by tony and is a combination of digital hash scores and relativity transformations using a system of systems approach. It has been applied to a range of issues including the common reporting scheme, the Panama Papers, weather prediction, law enforcement and emergency Services tasking and predicting modelling. Tony will cover how open source datasets from the ABS and UN/World Bank can be turned into data sequences to identify cohorts in a target population for strategic, operational and tactical activities.
10.00 AEDT — Dr Eugene Dubossarsky — Building Successful Data Analytics Capability
The key challanges in data analytics are not deeply technical. Rather, they are cultural, managerial and strategic. How to attract and develop the appropriate staff, how to manage the team effectively and how to recognise their success and convert it to value : these are the main challenges facing many current and aspiring executive sponsors, owners and managers of data analytics functions. This presentation addresses these challenges, and provides key reframing concepts for success.
10.30 AEDT — Keynote – David Skillicorn — Using language to understand mental state
Looking for bad actors and their bad actions is easier in datasets where it is hard for them to control how they appear, especially in comparison to normality. Language use is a particularly rich source of such data because we, as humans, have much less control than we think over our speech and writing. Our language leaks our mental state, but in ways we find difficult to perceive directly. I will illustrate how language usage can be leveraged to detect deception and fraud, intensity of jihadist language in forum postings, effectiveness of islamist propaganda, and success in winning U.S. presidential elections (in normal cycles). My talk will also cover other models of influence, interestingness, and gamification.
11.40 AEDT — Roundtable — All speakers
12.00 AEDT — End