#################################
#################################
Usage: detect-outliers [threshold] OR run [threshold] Detect outliers from and using two methods: k-means and COD (in this case, k-medoids-based). Output will be written to k-means-dataset1.dat, k-means-dataset2.dat, cod-dataset1.dat, and cod-dataset2.dat.
ex.:
$ ./detect-outliers Dataset1-outlier.arff Dataset2-outlier.arff
Alternatively:
$ java -cp target/classes edu.tufts.cs.ml.cluster.outlier.drivers.DetectOutliers [threshold]
#################################
#################################
Usage: cluster [out] Clusters features from using two initialization methods-- random and density-based-- and outputs the statistical results of each into [out] or, if [out] is not provided, the default file: kmeans.out.
ex.:
$ ./cluster segment-full.arff
Alternatively:
$ java -cp target/classes edu.tufts.cs.ml.cluster.drivers.ClusterWithKMeans [out]
#################################
#################################
Usage: active-learn Learns features from using two methods-- random and uncertainty sampling-- and outputs the results of each to their respective output files.
ex.:
$ ./active-learn train.arff random.out uncertainty.out
Alternatively:
$ java -cp target/classes edu.tufts.cs.ml.learning.active.drivers.ActiveLearn
Note: you can also run the active learner for 10 iterations at a time by invoking this class:
$ java -cp target/classes edu.tufts.cs.ml.learning.active.drivers.BatchActiveLearn
#################################
#################################
Usage: select-features Selects features from using three methods-- Filter, Wrapper, and my own combination-- and outputs the results of each to their respective output files.
ex.:
$ ./select-features train.arff filter.out wrapper.out own.out
Alternatively:
$ java -cp target/classes edu.tufts.cs.ml.features.drivers.SelectFeatures
#################################
#################################
Usage: validate [out] [--normalize] Validates using LOOCV and, if normalizing, outputs normalized data to [out].
ex.:
$ ./validate train.arff out.arff --normalize
Alternatively:
$ java -cp target/classes edu.tufts.cs.ml.validate.drivers.Validate [out] [--normalize]
#################################
#################################
Usage: preprocess Fills in missing values in and based on the data in and the method . Valid values for are: {MEAN, MEDIAN, MEAN_SAME_CLASS}
ex.:
$ ./preprocess train.arff test.arff output_train.arff output_test.arff MEAN
Alternatively:
$ java -cp target/classes edu.tufts.cs.ml.preprocess.drivers.ProcessMissingValues
#################################
#################################
Usage: classify [--normalize] Trains on TRAIN, classifies TEST, and writes results to OUT.
ex.:
$ ./classify train.arff test.arff output.arff --normalize
Alternatively:
$ java -cp target/classes edu.tufts.cs.ml.classify.drivers.ClassifyWithKNN [--normalize]
Kelly Moran kmoran@cs.tufts.edu