OVERVIEW
ADVANTAGES
»
PROCESS
APPLICATIONS
SCREEN SHOTS
 |
|
|
The Datascape Process is typically comprised of the four
steps illustrated in the diagram below. |
|
Quickly
Construct Models that Capture the Complex Behavior of your Data...
|
|
Preparation
Preparing
the data is the first step in creating a model. Each data point
consists of a series of inputs and an associated output. Datascape
will create a model of the output based on the relationships it
learns from the data points. Sources of data can include: test
and acquired data or data tables, output from a simulation.
|
Learning
Datascape
uses an iterative learning process to build a model from the data
points. You can even observe the evolution of your model during
the learning cycle.
Datascape provides the analyst with interactive:
- 3-D Surface Visualizations, with Input Controls for Additional
Dimensions
- 2-D Plotting Functions of Correlation and Model Response
- Quality of Fit Metrics
- Quantitative Ranking of Input Importance
- Visualization of Data Points, Including those that Don't
Follow the Trend
|
|
Validation
After learning, a model can be validated
by testing it with the verification data set developed in the
first step.
|
|
Deploying the Model
After creating a model, you can use it to
evaluate new inputs through the Datascape API, allowing you to:
- Create New Standalone Applications
- Link into Existing Applications
- Embed the Model into Real-Time Systems
The
Datascape API is currently implemented for the following environments:
- Microsoft C++ (DLL)
- SGI IRIX (DSO)
- RedHat Linux (shared library)
- HP-UX 64-bit Itanium (shared library)
- Embedded RTOS
Other platforms will be supported as required.
Datascape models can also be
used in Microsoft Excel spreadsheets via the Datascape/Excel Add-In.
|
|
|
|
