Geospatial Datathon - Exploring Geospatial Data Analytics
15 & 16 May 2021
Geospatial Datathon 2021
The WA Data Science Innovation Hub is pleased to announce the 2021 Geospatial Datathon delivered in partnership with Pawsey Supercomputing Centre, Department of Communities and the Australian Space Data Analytics Facility. The Datathon will bring together people working across a wide range of industries to explore the challenges, opportunities and innovations presented by geospatial data analytics.
Participants will work in multidisciplinary teams and apply geospatial data analysis tools and techniques to address real world problems presented by our industry partners. The sessions will kick off with a launch event on Friday 14th May.
Participants will need to attend in person in Perth, Western Australia.
Who should participate?
The event is open to GIS specialists wanting to connect with the broader geospatial community in WA, data scientists in other fields with an interest in upskilling, and data science students interested in geospatial data analysis.
People will be able to register as individuals or as a pre-formed team of individuals.
Interested participants will be able to attend a training session on Thursday 13th May to learn more about working with geospatial data. This session will be run by Dr Mortaza Rezae, Training Coordinator at the Australian Space Data Analysis Facility (ASDAF). The training will be organised into two sections:
- Discuss the fundamental concepts that define satellite imagery
- Assess the characteristics of satellite imagery
- Introduction to the structure of imagery data; and
- Choosing the best imagery to meet your project objectives.
Hands-on opportunity to explore and analyse high-resolution imagery data. Specifically, you will learn how to perform foundational data analysis using Numpy, Matplotlib, and Rasterio Python libraries, and apply these techniques to satellite data.
Participants can select their preference for the four proposed industry challenges:
CHALLENGE 1 – Spatial anonymisation: Place-based analysis has enormous potential to add to our knowledge, insights, and service delivery. The most significant utility can often come from sharing data across agencies. However, de-identification protocols often require aggregation up to a sufficiently large geospatial area to protect individual anonymity. Aggregation to any boundary introduces the modifiable areal unit problem. If these are in sparsely populated areas, data is suppressed for any aggregated line item under 5. For detailed on individuals in remote areas, we must aggregate up to very large spatial units, SA2. In addition, many of our populations of interests live in areas where spatial boundaries do not match standard spatial boundaries. Remote aboriginal communities are very poorly represented, even at the mesh block level of ABS data.
Given all these issues, what are the other options? How do we share information, keep personal information private, and maintain data quality that allows for meaningful analysis? Can we find a safe, effective alternative to suppression? Or must we accept a higher level of risk of re-identification?
CHALLENGE 2 – Incomplete data: Standard data sets often fail to capture information on specific groups of people. There are numerous reasons for this, but mainly they are challenging to reach or transient populations. These groups include international refugees, internally displaced persons, homeless and remote aboriginal communities. If we were dealing with demographic data, we might look to alternative sources to flesh out this data, for example, looking to service providers, schools, or NGO’s. In the case of GIS data, unless these secondary sources specifically capture location information, this method does not offer the same opportunity to fill gaps. Humanitarian OpenStreetMap has a worldwide network of community mappers working to capture refugee movements, and access to basic needs, amongst other things. Unfortunately, the same resources are not always available, nor necessarily appropriate for other groups.
How can we use other GIS data sources to better estimate the population and movements of underrepresented and transient people? Create a tool designed to solidify spatial data on your chosen target group.
CHALLENGE 3 – Data amalgamation for emergency response: The Department of Communities has state-wide emergency welfare authority. As such, we are responsible for providing emergency welfare support in the case of all-natural disasters, pandemics, mass casualty events, satellite re-entry, and any other events causing major social impact. Given this, we need a product that can pull together multiple live data sources that can then be deployed in house for analysis with our client data. Given the wide range of emergencies that we are responsible for responding to, and their complex nature, a tool like this needs to pull together live datasets that are often in incompatible formats, with different levels of accuracy, and currency. This is not just a problem of GIS data logistics; this also requires analysis and decision making on where datasets can best compliment and interact with each other to add value.
CHALLENGE 4 – Data amalgamation for rental stress analysis: In addition to the ‘fast moving’ welfare concerns of bushfires and cyclones, we are also responsible for long running welfare trends that will increase demand on our services. COVID-19 and the subsequent policy changes have created one of the most unprecedented, and complex challenges for this area. We are operating in an environment that has created an exception to almost all data trends. As policy responses come on and off-line we need to anticipate their impacts, and understand these in a location based way. One of the most pressing of these analytical challenges is understanding what impact the end of the rental freeze and eviction moratorium will have on the private rental market, particularly the availability of affordable rentals.
We need a product that brings together core logic and other real estate data to understand where the rental market is changing most rapidly, and map it against welfare response resources, such as the food relief network, emergency accommodation, public transport, and areas already seeing high rates of homelessness.
What will the Datathon involve?
Geospatial Datathon Launch Event
WADSIH will be hosting a launch event for the Datathon so participants can meet their team and find out which challenge they will be tackling over the Weekend Collaboration Sessions.
WHEN: Friday 14th May, 5.00 – 8.00pm WHERE: Technology Park Function Centre, Bentley
Food and drinks provided.
Weekend Collaboration Sessions
There will be two Weekend Collaboration Sessions running from 9.30am-5pm (AWST) on Saturday 15th May and Sunday 16 May.
You must be able to commit to attending all Weekend Collaboration Sessions in person at Technology Park, Bentley in order to participate in the Datathon.
Over the weekend sessions, each team will work together to develop an innovative solution to their challenge. At the end of the final session, each team will pitch their idea to a panel of judges who will then select the top three proposals.
Following the Datathon, WADSIH will provide further support to those interested in future work through linkages with relevant research or industry projects.
More information on the structure of the sessions and next steps for the top three proposals will be provided at a later date.
We will be holding a Geospatial Webinar on Tuesday 27 April 2021 to provide more information on the Datathon.
The webinar will be led by WADSIH with speakers from across industry presenting on the various applications of geospatial data analytics, and valuable insights into how WA is using geospatial data to solve real world problems.
Register for the Geospatial Webinar here
Express Your Interest
If you would like to take part in the Geospatial Datathon, please register your interest HERE
Should you have any questions regarding the Geospatial Datathon, please contact firstname.lastname@example.org