In the geographic information system (GIS) realm, visualization is different from analytics, but the distinction isn’t well-recognized. How do we clearly mark where one stops and the other begins? Is it a missed opportunity when an organization chooses to stick to simple data visualization on a map rather than performing location-based analytics?
Three months ago, my team and I started collaborating on a demonstration for a new Esri product coming out, called “Tracker for ArcGIS”. Basically, Tracker is a simple app that records the current location of any mobile device every few seconds and then periodically syncs the tracks back to ArcGIS Enterprise. Within no time, we collected gazillions of points of where team members had travelled. We put all these points on a map to visualize them. Great. But after a while, when we had had enough of looking at data and talking about these points, we asked ourselves: so, what?
It was actually a pretty boring demo but it got me thinking. Sure, we can visualize large datasets and symbolize them any way we want, but where do we go from there? What’s the next step after visualization?
So, we delved deeper. Once we analyzed those points together, we noticed some interesting things. For example, an aggregation of the speed recording of each of the hundreds of thousands of points collected revealed that when Sue drove on the Gardiner Expressway every day, her speed plummeted on a regular basis and then she consistently slowed down at one particular point. That’s a pattern.
We analyzed further by throwing in variables like changing the times of the day. We also observed other repetitive incidents. By tracking these patterns and trends, we could make certain predictions about Sue’s driving route. If she leaves at a particular time of the day on a specific route, she could reduce her travel time. Now, if she wanted to complete the task within a given timeframe, we could prescribe the time and route she needed to take. This is the power of analytics: moving beyond visualization.
Here lies the distinction between using GIS for visualization and analytics. While both take place using a map and seem closely related, they are not the same.
The popularity of business intelligence (BI) tools for dashboarding and visualization have, I think, blurred the distinction between visualization and analytics. To be clear, I think both good visualization and robust analytics are important; however, we have to clearly articulate and understand where one stops and the other begins.
When you see a neat dashboard with pie charts and maps, you might think it’s “analysis”, but it is not. Simply visualizing the results of an analysis that was done manually or with the help of other tools is not GIS analysis. In other words, people might look at a map that shows where a political party won most of the votes and refer to it as spatial analysis. But, it’s visualization. You can visualize raw data, or you can visualize the results of analysis, but the pie chart, the bar chart or the dots on the map are not in-and-of-themselves “analytics”.
In the non-spatial BI world, analysis is done using tables with scripts or formulas and then the results are visualized in charts. In the GIS world, there isn’t a clear distinction in that we do our analysis in the map, and then we visualize the results in the map. So, it’s easy to see where those not familiar with GIS could see simple maps as “spatial analysis”.
Hidden from view are the powerful spatial analysis tools we have at our disposal that can drive greater value and add a new perspective to simple map visualization––proximity, aggregation, hot spots, territory analysis––these tools tell us more about what is going on.
GIS needs to return to its roots which lie in analytics. In fact, GIS started as an analytical offshoot of mapping––extending digital maps to do analysis. While we need data and data management, and of course, maps serve that purpose exceedingly well, we must not stop there. If organizations want to understand data, drive deeper understanding of patterns and relationships and find answers, analytics is the way to go.
Even when we have access to the tools, we often don’t do enough predictive and prescriptive analyses inside of ArcGIS; we publish maps and leave it to the viewer to tackle the “so what?” between their ears. Of course, maps and charts are good for description. We look at a chart or a map and make a decision. However, what we need to do is leverage all of the analytical tools in GIS and quantify our results, follow them through statistically and mathematically so that we can make better, well-analyzed, more confident decisions.
The GIS industry has come a long way in offering location analytics technology. Consider Insights for ArcGIS that allows organizations to perform spatial analysis––geoenrichment of demographic data, aggregation and proximity analysis. In addition, Esri has teamed with Microsoft to expose geo-analytical tools inside of Power BI, which allow you to reach out to your Esri GIS and pull in the results of deep spatial analysis, and visualize the results in the platform of your choosing. In other words, you can simply use a map to do analysis.
We need to raise the value of GIS within our organizations. This means going beyond the custodial use of GIS to manage data about assets and publish “click-me” maps. This means applying GIS––The Science of Where––to the tough problems facing our communities and organizations. It’s not easy; in fact, it’s hard. It takes math. It takes us out of our comfort zone. And that’s OK.
This article originally appeared in the Fall 2018 issue of ArcNorth News.