12981 Rate this article:

Analytics vs. Analysis


The term “analytics” has become ubiquitous, popping up in all areas of conversation, from personal finance commercials to professional sports play-by-play, to our own disciplines of geospatial science, software and solutions. I hear the term “geospatial analytics” more and more, and here at Harris Geospatial we use the variation “ENVI analytics” to describe capability. However, if it is used without the proper context, it can come across as shorthand for “trust us, our software is hip and cool” instead of conveying meaningful information to your customer.

So what are geospatial analytics and how do they relate to analysis? Analysis, after all, is the human activity that we develop software to enable, enhance and improve. When deployed on the desktop, software such as ENVI provides a rich environment of tools and algorithms that the analyst uses to interrogate the data and apply the principles of remote sensing in order to perform measurements, derive information, and visualize results in an interpretable way. It is inherently interactive and iterative; requiring intimacy with the data that deepens knowledge of the problem set and improves the analysts’ domain knowledge, subject matter expertise, and tradecraft. Take pan-sharpening as an example; prior to the SPEAR workflow released in ENVI 3.X, fusing multi-spectral imagery with a higher resolution panchromatic image of the same scene was performed manually, with the analyst co-registering the data through an iterative and labor intensive process, selecting and applying the best color transform, and writing the results to file in order to inspect the product. It almost never happened right the first time, but the hands-on approach deepened the discipline of analysis for the operator by forcing intimacy with the data and imbuing practical experience with the process.

Analytics, by contrast, are a short circuit to the answer, the so-called “easy button”, designed to apply specific algorithms to specificdata to deliver an answer within a well-bounded and defined set of parameters.  In the pan-sharpening example, such an“analytic” would require the analyst to need only select the data, define the boundaries of the pan-sharpened output, and hit “go”.  Geospatial analytics, like the pan-sharpening task described above, are possible in the ENVI Services Engine (ESE) because we have moved the processing from the desktop to the cloud, decoupled the analytical engines from the user interface, and have recast the processing and exploitation tools organic to ENVI into more atomic level tasks. These tasks, literally called “ENVI Tasks,” have evolved from, and now replace, the ENVI_DOIT routines that were the traditional building blocks of our ENVI Application Programming Interface (API). This evolution is not only transformative to the technology, allowing us to process data on the cloud and return results to our web viewer or thin client, it is transformative the application of analysis by the ENVI user.  

This can be thought of in more than one way. For example, by providing the analyst with a palate of pre-defined geospatial analytics, such as Viewshed, Change Detection, Supervised/Unsupervised Classification, ENVI in the cloud frees the analyst from the laborious processing needed to define and implement these analytics. One observes that this removes operator intimacy with the data, but in the era of big data, this has become a necessary evil in many applications. Or, thought of in another way, this palate of analytics can be used to string together workflows unique to the problem set that are easily customized on the fly to process new data types or investigate new phenomenologies. This restores and even enhances intimacy with the data, an interesting concept in the era of big data. In the first paradigm, ENVI provides the tools that will “democratize” geospatial data for end-users in vertical markets underserved today by geospatial processing. The consumer doesn’t need to know what a line of sight analytic does, but only cares if his roof is capable of receiving enough solar radiance to make a photo-voltaic array worth the cost of investment – ENVI analytics makesthat possible. In the second paradigm,the geospatial analyst not only cares what a line of sight analytic does but wants to optimize it to account for seasonal growth of nearby crops to ensure full temporal coverage – ENVI analytics not only does that, but makes it possible for the data-intimate analyst to create complex and scalable analytical tools to solve more problems effectively and quickly. 

To answer the question posed in the title “analytics vs. analysis”, think of ENVI analytics as the distillation of the exploitation and processing tasks needed to support human analysis. Analytics are the atomic level tasks that are easily configurable and deployable, enabling the on going and accelerating migration of you geospatialenterprise from desktop to the cloud.