Interview with a Data Scientist


Dr Joseph Bailey is the Data Scientist working on the Tombolo Digital Connector. He has been involved in the development of the tool and its use cases for 21 months. He specialises in analytics, data visualisation and data science communication. He has over 7 years experience working on research and development of sustainable technologies and city-based challenges in academia, private industry and government supported organisations. This interview sheds light on how his experience and opinions of the innovation process Tombolo has taken.

What has been the most challenging part of this process?

I’d say that it was reasonably easy to identify the challenge – the fact that it’s hard to combine models and data from different sources and that being able to do that would unlock real value for cities. From my perspective, the hard part was deciding how to address the challenge. Even though we said that we would create a system for combining models and data there are many ways to do this.  Deciding on an approach was difficult and in the end, it was partially informed by what was possible given our circumstances (which I guess is often the case).

If we had a million people and unlimited resources then it may have been possible to create a system capable of combining world leading models from the realms of transportation, air quality, health and planning (Etc…). In reality however, the range of different types of models, the accessibility of those models, the data needed to populate them, and even the knowledge needed to operate them made this unrealistic. Not to mention the resources required to generate the system capable of combining these things.

Nonetheless, given these challenges we’ve still created a tool that is great at combining data from different sources and enabling the generation / combination of models from different data for different people – so we’ve made it a few rungs up the ladder.


What has been your best/most rewarding moment?

Working with a multidisciplinary team has been incredibly informative and Tombolo continues to require different skills from a huge range of individuals. However, I’m really looking forward to the open source release of Tombolo so I’m holding back my reward/ excitement for that.


From the aspect of a data scientist, how have you managed working with people who have limited experience in working with data?

My biggest flaw has been assuming that others have the same perspective as me and that certain ideas are implicit. It’s very easy to assume that other people are thinking similar things to you, especially when you spend most of your time with data scientists and software developers. As I learned through Tombolo, people aren’t thinking the same thing as you as often as you think.  In some circumstances, it took two or three conversations about the same thing for me to realise that my colleague was imagining something from a different perspective. As a result, I had to completely reframe my thoughts to incorporate their perspective on the concept. Doing this has been a challenge but it has also taught me that you need diversity to develop a robust tool / idea.


What has been the most useful method of communication when explaining the data process?

Metaphors have been incredibly useful – especially with respect to external communication. In particular expressing metaphors through visuals such as our recipe diagram of the Tombolo Digital Connector.

Internally, I’d say that demonstrating ideas through use cases or even tool prototypes has been the most useful. Sometimes verbal explanations don’t suffice and actually doing something or making something to explain a concept is worth it. In the Tombolo project, some of our small demonstration tools have become useful ways of explaining the value of the project.


What are the pitfalls of undertaking a data tool innovation project, and how do you avoid/deal with these?

There’s always a chance that someone has already done what you’re doing, or is currently doing it. This is particularly true when you’re trying to develop an open source tool. There are lots of private industries that have the capacity to develop great tools quickly. However, the value they enable is often locked behind a paywall and cannot be realised by cities. On the plus side, even if people do generate tools that do similar things, we (or others) can take the best bits of each and make something even better resulting in a greater good.


If you were to do this project again, what would you do differently?

I’d try and involve more people as soon as possible to really ‘hype’ the project. Tombolo has a lot of potential and making as many people aware of the project as possible would have been great. I think as a team we’re aware of this and we’re taking steps to better document our work and share it more widely.  Increased sharing may also have increased our opportunities to test our ideas / tools as soon as possible. Sooner and more frequent testing may have enabled faster learning which is always good. Finally, for the project to be really responsive to feedback I’d include more development capacity. Being able to implement the ideas / feedback from testers quickly would really allow us to ‘pivot’ the development of Tombolo in light of our learnings.


What has been your biggest learning from another discipline?

No matter how good your ideas is, you still need to sell it. We need excellent communications and materials around our projects and we need to demonstrate their value to others clearly. Even if your product is not completed, you need to enable others to see the value in the vision.


When creating a tool, do you become attached to it? Does this limit the innovation of it?

It is definitely possible to become attached to your ideas. This is why it’s important to have diverse collaborators, opportunities for feedback and an acknowledgement that constructive criticism leads to learning. I think it’s important to instil the idea that for a team to make the best possible tool (for a particular challenge), they have to be willing to let sunk costs be sunk costs and use them as learning. I.e. don’t be afraid to drop what you have and start fresh with new knowledge.

Leave a Reply