This page contains most of the matlab scripts that I’m using for my work. I normally publish them in the posts on the right side and present them on this static page that lists them for easy access. The listing is in reverse chronological order, there are also comments inside the m-files regarding script requirements and possible dependencies.
Matlab with Sammon’s mapping
- sammon-experiments-0.m, performs Sammon’s mapping and generates the figures in http://blog.georgruss.de/?p=132.
- sammon-experiments-1.m, improved version of the above script.
- makemovie.m, script to be used in conjunction with the above Sammon’s mapping script; colors the data points where a variable is higher than a certain threshold and generates a movie from that to visually find the best split.
- sammon-experiments-2.m, another improved version of the above script. Now it generates the figures in the MLDM2009 paper proposal and can also create movies on Sammon’s mapping maps with appropriate coloring of the map to find out where to split the attributes. This doesn’t always work, but may help in gaining insight into the data. Once the paper has been accepted for the conference, I’ll post it here.
Matlab with SVMTorch
- SVM regression, first ideas and experiments, description at http://blog.georgruss.de/?p=94
- SVM regression, updated cross validation functions, some comments at http://blog.georgruss.de/?p=97
- SVM regression only, for figuring out SVM parameters, fully commented, purpose description at http://blog.georgruss.de/?p=98
- SVM regression vs. MLP, data not normalized, comparison of neural network vs. support vector regression on specific data set, see http://blog.georgruss.de/?p=101
- SVM regression vs. MLP, normalized data, same as above, with normalization (which reverses the result), as described in http://blog.georgruss.de/?p=108. This script happens to be the same as the one below, since I forgot to rename them whilst copying to the public_html dir. However, it should be easy to comment out anything that contains the letter sequence „rbf“.
- SVM vs. MLP vs. RBF, described in http://blog.georgruss.de/?p=109.
Matlab with neural network toolbox
- agricult_publication_icdm2008.m, Figures for ICDM2008 conference (described at http://blog.georgruss.de/?p=62)
- RBF parameters, experimental, in conjunction with SVM models above, described in http://blog.georgruss.de/?p=112
Matlab, model comparison
- SVM vs. MLP vs. RBF vs. RegTree vs. LinReg vs. NaivePredictor (outer script, calls the one below, description in http://blog.georgruss.de/?p=162)
- updated modelcomparison.m (inner script, replaces earlier version, description in http://blog.georgruss.de/?p=162)
- Wrapper script for model comparison, used in http://blog.georgruss.de/?p=145
- model comparison function (script) for use with the above wrapper
- Regression Tree Script, used in http://blog.georgruss.de/?p=118
- SVM, MLP, RBF, RegTree compared, results in http://blog.georgruss.de/?p=122
(more entries will be updated over time not follow, since I switched to R for my computing stuff, see R scripts)