With SOM Analyzer you can create, train and test the SOMs against data sets the SOM has never seen. You can analyze any data that you can think of and find correllations and properties you never know existed. SOM analyzer helps you all the way in your quest by providing you with various data manipulation and visualization features.
The main functionality of the SOM Analyzer can be listed as follows:
NEW!
With the release of the new version 1.02, a Result Data Export feature has been added. It lets you export the BMU and cluster classification results (found by the SOM to your input data) to a CSV (Comma Separated Values) file format. You can then easily import that file to your favourite spreadsheet program for further analysis. To understand more, check the user guide.
Powerful wizard providing you with the abstraction level to using SOMs the way you see fit
Fully automatic mode chooses all SOM parameters for you. The only thing left for you to do is selecting the data the SOM processes.
In semi automatic mode, it is sufficient to have a rough idea on what you are doing with the SOM. After this and after selecting the data for the SOM, you are good to go.
Manual mode lets you choose all parameters by hand and this way allows you full control over the SOM you are creating.
Iterator is something that you can leave running for as long as you want and its purpose is to help you create suitable maps. By iterating through a number of parameters, the iterator creates several SOMs in an attempt to find the SOM to suits your data the best. After iterator run completes, you are presented with a list of SOMs created during the process. Using the SOM quality metrics calculated, you can then try SOMs to see if it is ok. If you so choose, you can come back to the iterator list any time you want to have a go at another SOM.
As SOMs are visual by their very nature, the several visualization methods provided let you choose the way you want to see your map. The supported visualization methods are, Umatrix, Parameter level, Histogram and Cluster. SOM Analyzer is capable of showing any number of visulizations on screen at the same time. It does this by chopping the screen into segments where each visualization is shown.
Support for MySQL databases. By installing the MySQL Connector driver from here, you can start feeding SOMs with data that comes from your MySQL database. If you do not need a database, you can always have SOM Analyzer process the data in files.
Labelling lets you give meaning to either map areas or data columns so that your SOM is more intuitive to use.
Examining data could not be easier. Just click any of the map units to see its data on the left pane. Alternatively, you can just hover your mous on top and a tool tip appears showing you the date. In addition, there is a context-sensitive menu for each visualization method to fine tune the appearance.
Automatic Clustering lets you dive into the map data by extracting the number of clusters you want. You can either let the clustering determine the clusters automatically or you can define the clustering parameters yourself. In cluster visualization, clusters can be labelled to give them meaning.
Offline and online testing. You can perform SOM testing whether you already have all the data you need in offline testing mode. If new data keeps coming all the time, you can fetch it real-time from database.
In testing, it is useful to see the map units the operating point has traversed. To this end, the SOM Analyzer features a trajectory functionality that plots the path the operating point passes.
Find BMU feature shows you a dialog where you can enter the data components you want. Based on the data, the Best-Matching Unit (BMU) is found.
Loading and saving maps is perhaps too obvious to mention. You can save maps you create and upon loading, restore the state with all the visualization exactly as they were.
Tailor SOM Analyzer to best suit your needs. You can set & save various options and colors so that you get a working environment suited to your taste.
The comprehensive on-line user guide is accessible directly from SOM Analyzer.
Take a look at the screenshots and the attached descriptions for an illustration on how these features look in action!
Supported Platforms
SOM Analyzer should support all Windows releases since and including Windows XP. It has been tested on Windows XP, Windows Vista, Windows 7, Windows 8 & 8.1 and Windows 10.