For all the talk about analytics these days, there has been little mention of one of the most powerful techniques for analyzing data: location intelligence.
It's been said that 80% of all transactions embed a location. A sale happens in a store; a call connects people in two places; a deposit happens in a branch; and so on. When we plot objects on a map, including business transactions and metrics, we can see critical patterns with a quick glance. And if we explore relationships among spatial objects imbued with business data, we can analyze data in novel ways that help us make smarter decisions more quickly.
For instance, a location intelligence system might enable a retail analyst working on a marketing campaign to identify the number of high-income families with children who live within a 15-minute drive of a store. An insurance company can assess its risk exposure from policy holders who live in a flood plain or within the path of a projected hurricane. A sales manager can visually track the performance of sales territories by products, channels, and other dimensions.
Geographic Information Systems. Location intelligence is not new. It originated with cartographers and mapmakers in the 19th and 20th century and went digital in the 1980s. Companies, such as Esri, MapInfo, and Intergraph, offer geographic information systems (GIS) which are designed to capture, store, manipulate, analyze, manage, and present all types of geographically referenced data. If this sounds similar to business intelligence, it is.
Unfortunately, GIS have evolved independently from BI systems. Even though both groups analyze and visualize data to help business users make smarter decisions, there has been little cross-pollination between the groups and little, if any, data exchange between systems. This is unfortunate since GIS analysts need business data to provide context to spatial objects they define, and BI users benefit tremendously from spatial views of business data.
Convergence of GIS and BI
However, many people now recognize the value of converging GIS and BI systems. This is partly due to the rise in popularity of Google Maps, Google Earth, global positioning systems, and spatially-aware mobile applications that leverage location as a key enabling feature. These consumer applications are cultivating a new generation of users who expect spatial data to be a key component of any information delivery system. And commercial organizations are jumping on board, led by industries that have been early adopters of GIS, including utilities, public safety, oil and gas, transportation, insurance, government, and retail.
The range of spatially-enabled BI applications are endless and powerful. "When you put location intelligence in front of someone who has never seen it before, it's like a bic light to a caveman," says Steve Trammel, head of corporate alliances and IT marketing at ESRI.
Imagine this: an operations manager at an oil refinery will soon be able to walk around a facility and view alerts based on his proximity to under-performing processing units. His mobile device shows a map that depicts the operating performance of all processing units based on his current location. This enables him to view and troubleshoot problems first-hand rather than being tethered to a remote control room. (See figure 1.)
Figure 1. Mobile Location Intelligence.
A spatially-aware mobile BI application configured by Transpara for an oil refinery in Europe. Transpara is a mobile BI vendor that recently announced integration with Google Maps.
GIS Features. Unlike BI systems, GIS specialize in storing and manipulating spatial data, which consists of points, lines, and polygons. A line is simply the intersection of two points, and a polygon is the intersection of three or more points. Each point or object can be imbued with various properties or rules that govern its behavior. For example, a road (i.e., a line) has a surface condition and a speed limit, and the only points that can be located in the middle of the road are traffic lights. In many ways, a GIS is like computer-aided design (CAD) software for spatial applications.
Most spatial data is represented as a series of X/Y coordinates that can be plotted to a map. The most common coordinate system is latitude and longitude, which enables mapmakers to plot objects on geographical maps. But GIS developers can create maps of just about anything, from the inside of a submarine or office building to a geothermal well or cityscape. Spatial engines can then run complex calculations against coordinate data to determine relationships among spatial objects, such as the driving distance between two cities or the shadows that a proposed skyscraper cast on surrounding buildings.
Approaches for Integrating GIS and BI
There are two general options for integrating GIS and BI systems: 1) integrate business data within GIS systems and 2) integrate GIS functionality within BI systems. GIS administrators already do the former when creating maps but their applications are very specialized. Moreover, most companies only purchase a handful of GIS licenses, which are expensive, and the tools are too complex to use for general business users.
The more promising approach, then, is to integrate GIS functionality into BI tools, which have a broader audience. There are several ways to do this, which vary greatly by level of GIS functionality supported.
- BI Map Templates. Most BI tools come with several standard map images, such as a global view with country boundaries or a North American view with state boundaries. A report designer can place a map in a report, link it to a standard "geography" dimension in the data (e.g. "state" field), and assign a metric to govern the shading of boundaries. For example, a report might contain a color-coded map of the U.S. that shows sales by state. This is the most elementary form of GIS-BI integration since these out-of-the box map templates are not interactive.
- GIS Mashups. GIS mashups are similar to BI mashups above but go a step further because they integrate with a full-featured GIS server, either on premise or via a Web service. Here, a BI tool embeds a special GIS connector that integrates with a mapping server and gives the report developer a point-and-click interface to integrate interactive maps with reports and dashboards. In this approach, the end-user gains additional functionality, such as the ability to interact with custom maps created by inhouse GIS specialists and "lasso" features on a map and use those selections to query or filter other objects in a report or dashboard. Some vendors, such as Information Builders and MicroStrategy built custom interfaces to GIS products, while other vendors, such as IBM Cognos and SAP BusinessObjects, embed third party software connectors (e.g., SpotOn and APOS respectively.)
- GIS-enabled Databases. Although GIS function like object-relational databases, they store data in relational format. Thus, there is no reason that companies can't store spatial data in a data warehouse or data mart and make it available to all users and applications that need it. Many relational databases, such as Oracle, IBM DB2, Netezza, and Teradata, support spatial data types and SQL extensions for querying spatial data. Here both BI systems and GIS can access the same spatial data set, providing economies of scale, greater data consistency, and broader adoption of location intelligence functionality. However, you will still need a map server for spatial presentation.
As visual analysis in all shapes and forms begins to permeate the world of BI, it's important to begin thinking about how to augment your reports and dashboards with location intelligence. Here are a few recommendations to get you started:
- Identify BI applications where location intelligence could accelerate user consumption of information and enhance their understanding of underlying trends and patterns.
- Explore GIS capabilities of your BI and data warehousing vendor to see if they can support the types of spatial applications you have in mind.
- Identify GIS applications that already exist in your organization and get to know the people who run them.
- Investigate Web-based mapping services from GIS vendors as well as Google and Bing since this obviates the need for an inhouse GIS.
- Start simply, by using existing geography fields in your data (e.g., state, county, and zip) to shade the respective boundaries in a baseline map based on aggregated metric data
- Combine spatial and business data in a single location, preferably your data warehouse so you can deliver spatially-enabled insights to all business users.
- Geocode business data, including customer records, metrics, and other objects, that you might want to display on a map.
Location intelligence is not new but it should be a key element in any analytics strategy. Adding location intelligence to BI applications not only makes them visually rich, but surfaces patterns and trends not easily discerned in tables and charts.
Posted September 19, 2011 2:51 PM
Permalink | No Comments |