View from the North: Analytics – The Key to Transformation
Geospatial analysis can add to an organization’s business intelligence. Chris North explains in his regular column for ArcNorth News - how to use the multiple analytical insights from GIS.
I recently watched an interesting video of a session from the Esri User Conference 2017 in San Diego. The theme for the session was UPS’ use of GIS and was presented by Jack Levis, who is a senior director at UPS. I was quite intrigued to see how they used GIS – specifically analytics to optimize their operations. Now, the video is about 35 minutes. What really jumped out at me was Jack’s perspective on analytics (which, if you’re impatient like me, starts around the four-minute mark of the video).
Jack described four types of analytics: descriptive, diagnostic, predictive and prescriptive. This really got me thinking about how analytics has the power to help transform GIS, and more importantly — how analytics can help GIS transform our organizations.
The Transformational Value of Analytics. Source: ibm.com
Types of Analytics
The four types of analytics Jack talks about in the video are well documented elsewhere, but let’s take a moment and put them into the context of a GIS.
Descriptive analytics is what GIS has historically been all about. Most of us use this capability of GIS today – we make a map – driven by our data – to show where things are (or were). Perhaps we generate heat maps of our 311 service requests to see where the greatest number of calls are coming from. In a sense, this is the fundamental level of analytics we all do in a GIS. It’s basically saying, “Hey look, we have a lot of potholes here.”
Diagnostic analytics is not done quite as often, but is still commonplace in GIS. With diagnostic analytics, we add more data layers into the GIS, and see if there is some sort of quantifiable relationship between the two datasets. Now, we’re saying, “Hey look, we get more potholes on older roads than newer roads.”
Predictive analytics goes a step further. Based on the diagnoses of the relationships in our data, we attempt to extrapolate into the future (if we have time elements) or extrapolate into other geographic areas, where we can anticipate the same phenomena to happen. Now we are saying, “Hey look, since these roads are going to get older too, we can expect more potholes over here in three years.”
Before I get into the fourth type of analytics, it’s worth noting that the vast majority of GIS implementations I have seen never get beyond these three levels of analytics. Don’t get me wrong – there is value in these levels of analysis. Think about the statements I made about potholes after each type of analytics. To me, they all leave me asking the same question: “So what?”
Prescriptive analytics is going the next step to reveal what actions should be taken to either correct a negative trend or capitalize on a positive trend. With prescriptive analytics, you are testing different hypotheses (that come out of your diagnostic and predictive analyses) to determine which is the better course of action based on some metric (usually cost). It’s a lot more complex but brings greater value to the organization. With prescriptive analytics, you can say, “Hey look, since we are going to get more potholes over here in three years, it will cost us $X to wait and fill them in, but it will cost $Y to repave those roads two years from now. So, we should go with this option.” Yes, I’m simplifying for illustrative purposes.
As argued by Jack Levis and others , predictive analytics – delivered largely in a self-service paradigm across the organization - is where we all need to go to truly transform our organizations. The full value of GIS to an organization is the ability of GIS tools to empower all levels of analytics — from descriptive to diagnostic to predictive and finally, prescriptive.
But there’s more to the story. It's certainly true that the ability to provide these multiple levels of analytics is not unique to GIS. There are all kinds of business intelligence (BI) tools out there. But for most of these tools, maps are little more than a new chart type. It's also true that non-spatial BI tools can provide perspectives that may not be obvious from a purely spatial perspective.
I think that it is by adding the power of spatial analysis to existing BI workflows that we can help transform our organizations. We are already empowered with the GIS tools to tap into this resource; let’s begin utilizing these tools to their full potential so we can actuate real transformation in our organizations.