Envision is a novel approach to scientific visualization that enables scientists to see and understand data in new ways. Envision was developed by Dr. Michael A. Gilbert, a scientist at the University of California, Berkeley, and Dr. Christopher D. Wickett, a postdoctoral fellow at UC Berkeley. Envision is based on the idea that visualizing data can be more effective than traditional methods of data analysis, such as statistical methods or numerical simulation.
Envision has been used to study a wide range of problems in science and engineering, including the spread of diseases, the formation of galaxies, and the dynamics of fluids. Envision has also been used to create artworks and educational materials. The software is available for free online (http://www2.envizio.net/) and has been downloaded over 100,000 times since its release in 2009.
How does Envision work?
Envision uses an approach called “multi-dimensional scaling” (MDS) to visualize data sets with many variables. MDS is a mathematical technique that allows one to represent high-dimensional data sets in two or three dimensions without losing important information about relationships among variables.” In other words, MDS helps you turn complex data sets into simple pictures that you can easily understand.”
To use Envision, you first need to load your data set into the software. Envision supports several different file formats, including comma-separated values (CSV) files and tab-delimited text files.” Once your data is loaded into Envision, you can select which variables you want to visualize and how you want them represented.” For example, you could choose to represent each variable as a point in space or as a line connecting two points.”
If your dataset has many variables (>20), it may be helpful to first reduced the dimensionality using Principle Component Analysis (PCA). PCA finds new combinations of your existing variables that are uncorrelated with each other (“orthogonal”) and ordered by how much variance they explain in your dataset.” These new variables (“components”) can then be visualized just like any other variable in En vision.”