In the second article of this three-part series, we'll explore the landscape of the location intelligence (LI) world today, the technologies that are shaping it and the use cases that inhabit it.
Location Intelligence Landscape
The business intelligence
(BI) atmosphere today is rich in technology-driven innovation that has ignited a global warming trend for location intelligence within the BI
community. Location intelligence icecaps at the ends of the BI world are rapidly shrinking and affecting the surface currents of both business processes and business intelligence. Just as three-quarters of the earth is water, three-quarters of the corporate data world is composed of location data of some kind, from addresses to coordinates supplied by field workers via GPS devices to the locational data captured by billions of personal devices, such as smartphones and tablets.
There are also continents of spatial data in the public sector and in industries such as utilities, petroleum, transportation, natural resources and defense that have been shaped by spatial visualization and analysis for more than 30 years. Even here, however, vast expanses are uncharted and uninhabited in terms of location intelligence. Most exploration has been focused on mapping the terrain, not analyzing it, and full exploitation of this resource is still beyond the grasp of most BI analysts and consumers, known only to a relatively small group of trained geospatial professionals. Still, the analysis that has been done by these specialists is quite remarkable, although virtually unknown to the broader BI community: finding energy sources, saving civilian and military lives, keeping the lights on, growing forests, and keeping supply chains moving.
Location Intelligence Technologies
Perhaps the most important aspect of LI technologies to understand is the DNA they share with mainstream IT/ BI technologies and usage patterns. From the databases managed by IT to the tablets in the hands of CEOs, LI technologies mirror mainstream technologies. We'll stay on the beaten path here and not venture down the many side trails that lead to some interesting places but are not well visited at this point. Our main goal is to show the parallels between the BI world and the LI world, hopefully piercing some of the fog that surrounds the latter. Spatial Databases
A good place to start the discussion on LI technologies is the place where all data naturally comes together in the enterprise: the database. Although there are many proprietary spatial file formats in wide use, the most serious and interesting LI applications use a database to store and manage spatial data in the same manner as other data, such as numbers and text. Spatial data types and functions are simply another data type with associated manipulation functions added to the database.Spatial ETL/Data Quality Tools
Nearly all leading databases these days have spatial data types and functions available for free or as options including Oracle, DB2, and SQL Server. You create and do things with them using SQL as you do with other data types. After a bit of practice, a BI analyst should have little trouble doing some simple spatial queries and analysis. Spatial data types and functions are also appearing in pure data warehouse products such and Teradata and Netezza.
Location Intelligence Applications
There are a wide variety of spatial data residing explicitly in files and databases or hidden in non-structured text such as addresses or place names in documents or other media. Spatial data extraction, consolidation, and quality is playing an increasingly important role as acquisitions often leave companies with multiple location intelligence vendors and applications. As in the BI world, spatial ETL
and data quality
products can help access spatial data, clean it up, bring it together, keep it in sync, and inject it into business processes and BI applications. For BI professionals there should be no surprises here: spatial ETL tools reflect familiar capabilities and usage patterns found in similar ETL tools used for BI.
Desktop applications. As in business intelligence, desktop LI applications combine SQL access with business logic for a wide range of applications across nearly every public and commercial market. Some of these applications are general purpose and suitable for all industries while others focus on particular industries or applications, such as retail site selection or exploration for oil companies. Some primarily access spatial data for visualization; analysis and interpretation is left to the user. Others take full advantage of spatial relationships in the data such as proximity, containment, or connectivity and use them in complex models such as those used to optimize a network of retail locations or the sequence of drilling oil wells. Finally, they vary widely in their capabilities for creating and managing spatial data, which can have a huge impact on data quality.
Spatial servers. As in the BI world, desktop LI software often works with server-based software for delivering location intelligence to a broader enterprise community. These servers power map-centric LI applications as well as embed them into traditional BI tools or directly into operational business applications, bringing maps and spatial analysis to transactional business systems. Maps and spatial analysis might be visibly part of the user interface or might run in the background performing spatial calculations or optimizations.
Web applications. The LI landscape, like the BI landscape, has undergone a major reshaping towards web deployments. There are thousands of web applications with some form of mapping in them primarily targeted at the consumer, but a growing number are targeted at the business. These applications generally leverage Web 2.0 technologies for building highly visual and interactive maps. A great improvement over the static maps of the early Web, the interactive maps of Web 2.0 bring viewing, selecting and filtering spatial data to the Web, but generally fall short when it comes to spatial analytics, although this is rapidly changing.
Mobile applications. Map-centric applications for mobile devices have been around for over a decade for operational field workers who need maps to navigate or to collect geo-referenced data in the field. Much less has been done to take location intelligence to the field, but this is also true for BI applications, which are only recently being re-written to bring reporting to the legions of executives and managers who live by their flight schedules, smartphones and tablets. Here, the focus of location intelligence is again visualization, selection and filtering: Few location analytics, beyond simple things like routing, have moved to the field within BI applications.
Cloud applications. Perhaps the most influential technology for accelerating the adoption of location intelligence into mainstreamed BI is the Cloud. Already available from both small and large vendors, location intelligence resources in the cloud are a preferred option for bringing map visualization and analysis not only to the traditional industries that have long used spatial data and analysis but also to other industries where location intelligence has been largely ignored. As cloud security holes become solidly plugged, businesses will begin to move customer, sales, and other sensitive data on facility, supplier, and market locations beyond the firewall and into the Cloud. By this time, location intelligence vendors will have re-tooled their most complex spatial data creation, visualization, and analysis capabilities for the cloud and location analytics will assume a much larger role in the decision making process both within BI applications and business processes requiring real time spatial data and analytic processing.
The location intelligence application landscape should look familiar to the BI community as its features, usage patterns, and evolution are similar. With a little exploration and a quick course on the native language, the BI professional should have little trouble navigating the world of location intelligence.
Location Intelligence Use Cases
Another parallel between the BI world and the LI world is the nature of their applications. Location intelligence applications support strategic, managerial, and operational decision making such as where to deploy resources (near transportation or inputs to production) or how to employ them most efficiently (optimized routing). All involve decisions with important spatial components.
Location intelligence applications map to generally recognized categories of business intelligence. For brevity, we'll bucket them into broad categories of descriptive, predictive and prescriptive location intelligence.
Descriptive. Descriptive location intelligence generally replaces the tabular, pie and bar charts that show things like sales by region with geometric shapes representing countries, states or territories. Those who are geographically challenged may find some new insight based on spatial relationships in the data such as adjacency or proximity, or they may want to drill into more spatial detail or zoom out to see a broader pattern. This is the prevalent use case for location intelligence today. Visualization of the geographic dimension in standard reports and dashboards, and some limited ad-hoc query and analysis, is generally available in most BI tools today and in many map-centric LI applications running on desktops or the Web. Most BI applications fall short on features such as pixel-perfect cartography for hard copy or Web maps, or advanced spatial analysis.
This category of location intelligence also includes spatial statistics and data mining used for exploring data (finding outliers – bad addresses), measuring the geographic distribution of data (its center or compactness – nice to know for retail site selection), finding where something is likely to occur (minerals or hurricanes) or finding clusters of similar things (best customers).
Predictive. Modeling using spatial mathematics, spatial regression and other techniques helps with correlation analysis, trend analysis, and forecasting. Like other models, spatial models use a few variables to create a simplified, manageable view of the world to help solve problems or broaden understanding of how location impacts a decision, either directly (by changing some spatial variable) or indirectly (by viewing and analyzing the outcome within a larger geographic context outside of the model).
Spatial models can be quite simple, using only one data source and one spatial function, or quite complex, using many spatial data sources and multiple spatial functions. They can be applied to a wide range of problems such as those concerned with rating geographic areas (Where can I find oil or customers?) or making predictions of what might happen in an area (Will a natural or man-made disaster impact my supply chain or my insurance claims exposure?). Spatial models use historic and real-time data to help managers at all levels make better decisions.
Prescriptive. Knowing the lay of the land (descriptive location intelligence) and what may happen there (predictive location intelligence) paves the way for prescriptive location intelligence, in which decisions are assisted or prescribed by spatial relationships or models. Call center staff may be told to approve or decline a prospective policy based upon its location (e.g., proximity to a hazard). The location of a network of stores or the routing of service technicians may be optimized. Optimization in real time, such as for weather or traffic, brings even more value to the process. Optimization of the use of resources is the ultimate goal of both business intelligence and location intelligence.
We've taken a 30,000-foot flyover of the world of location intelligence and consequently missed a lot of detail about its features, but by now you should have a general comfort level that the LI world is very similar to the BI world and safe for further exploration by the BI professional. In the third and concluding part of this series, we will speculate on the future of location intelligence.
Recent articles by Steve Benner