Summary of Eonfusion's analysis capability
In this month's newsletter and the next we are presenting a number of case studies where Eonfusion is used for summary data analysis. The article will provide some context for those examples by explaining the various methods for performing summary analysis in Eonfusion and for using Eonfusion in conjunction with other tools.
Our case study examples use Eonfusion version 1.2, but most require prototype operators that will not become publicly available until version 2.0. Eonfusion V2.0 is due for release in late September, so all Eonfusion users will soon have access to these operators and methods.
The mechanics of data analysis involving Eonfusion can be separated into four general categories:
- Data analysed in Eonfusion using the available operators.
- Data analysed in Eonfusion using add-in operators, coded in the Eonfusion Application Programming Interface (API).
- Data exported from Eonfusion, analysed using external software, and re-imported.
- Data analysed by methods in external software that is accessed directly by Eonfusion.
Eonfusion's analysis capabilities are growing via the addition of new operators and methods. The Puma Tracking case study in this newsletter uses a density estimation method that counts the number of track vertices occurring within a grid of squares. The resulting density grid can then be overlaid and visualized with the raw track data, as illustrated in the image below.
This method was added to Eonfusion by simply enabling the creation of the necessary grid squares. From there, Eonfusion's attribute transfer and manipulation capabilities were used to generate vertex density values for each grid square. A similar approach can be used to generate Kernel Density Estimation (KDE) and other similar types of summary analysis.
While these analyses require some time and effort to build in Eonfusion's dataflow environment, once built they are very easily exchanged and distributed by simply exporting the relevant section of the dataflow. When version 2.0 is released we will distribute an example dataflow featuring this density estimation method.
The Eonfusion API will allow suitably skilled users to write add-in analysis operators for Eonfusion. The scope and capability of add-in operators is limited only by the capability of the programmer and the data structures that are handled by Eonfusion.
While not all users will have the programming skills required to write their own add-in analysis operators, they will be able to use add-in operators provided by other users and by us at Myriax. This ability to share and distribute add-in analysis operators will lead to explosive growth in the number of analyses that are available in Eonfusion.
However, even with these capabilities within Eonfusion the use of external software is sometimes preferable. This is often the case when there are established methods for data analysis via programs such as R or Matlab.
The Dingo tracking example in the Applications area of the website illustrates integration with the statistical analysis package "R" to generate a predictive analysis of time spent. Likewise the image below shows a time-spent surface generated in R from the underlying lobster track data.
Until Eonfusion version V2.0 these external analyses could only be performed by exporting data, analysing it using other software and re-importing the results.
However direct access to other software methods will be available via Eonfusion's API in version 2.0. Contact Eonfusion Support if you are interested to know more about integrating Eonfusion directly with other software.
