R scripts for exploratory data analysis using SOM. Frequently, I start working on a dataset which I know little about (in terms of noise, feature relevance, natural groupings) and need something to give me a 'picture' of how the data behaves.
That is why I created these scripts that can take a data file, normalize it, check for data inconsistencies and train an SOM. They also plot component plane visualizations, u-matrix and hit histograms. Hopefully I will be able to get it to a state where I can do as much as I can do with Matlab som_toolbox and even more.
Start R and change to the current directory. Run the script
test1.R
Which will take the scripts through both labeled and unlabeled data scenarios using iris dataset.
Should run on platforms that has R installed. kohonen and colorspace packages must be installed.