This project introduces an integrated approach to multivariate clustering and geovisualization ( in Java ), which builds upon the synergy of multiple computational and visual methods. Its unique strength is evident in several aspects. First, by leveraging the power of computational methods it can handle larger datasets and more variables than would be possible with visual methods alone. Second, it effectively synthesizes different perspective information to enable an overview of complex patterns across multiple spaces.
Third, its component-based design provides flexibility to extend the system by adding new components or replacing current components. Through the static linking and user interactions, complex relationships can be perceived and understood easily. However, there is a limitation on the color configuration — it is not always possible to assign a cognitively meaningful color to each cluster. The color surface must be smooth and continuous, meaning that the user may not always be able to assign a desirable color to a cluster, for example, a red color to represent a warming trend and a blue color to represent a cooling trend.
This project allows the user to rotate or flip the two-dimensional color surface to find a reasonable match between clusters and colors. Another potentially confusing factor is that when the user interactively changes the size of the SOM , the resultant clusters are different, and thus the multivariate meaning of colors will change. In other words, the same color may represent two different groups of data items for different runs of the SOM.
Advantages and Disadvantages:
- Multiple simultaneous perspectives.
- Ability to comprehend large amounts of data .
- Reduction in search time through visualization.
- Provides a better understanding of complex data sets.
- Reveal relationships and properties through visual perception.
- It is difficult to understand how much the vectors from one or other class correspond to a cell.
- We do not know how many vectors are from of the same class or of different class, and what their proportions are.
Download M.Tech Project on Integrated Approach To Multivariate Analysis And Geo-visualization Project Report