heri-eval(1) evaluate classification algorithm


heri-eval [OPTIONS] dataset [-- SVM_TRAIN_OPTIONS]


heri-eval runs training algorithm on dataset and then evaluate it using testing set, specified by option -e. Alternatively, cross-validation is run, if option -n was applied. If cross-validation is used, training and testing on different folds are run in parallel, thus utilizing available CPUs.


-h, --help
Display help information.
Enable output of per-fold statistics. See -Mf.
-n N
N-fold cross validation mode (mandatory option).
-t T
T*N-fold cross validation mode (1 by default).
-e testing set
Sets the testing dataset.
-o filename
Save results from testing sets to the specified file.

Format: golden_class result_class [score]

-O filename
Save incorrectly classified objects to the specified file.

Format: #object_number: golden_class result_class [score])

-m filename
Save confusion matrix to the specified file.

Format: frequency : golden_class result_class

-p opts
Pass the specified opts to heri-stat(1)
-M chars
Sets the output mode where chars are: t --- output total statistics, f --- output per-fold statistics, c --- output cross-fold statistics. The default is ``-M tc''.
-S seed
Pass the specified seed to heri-split(1).
Keep temporary directory after exiting.
Turn on the debugging mode, implies -K.


Training utility, e.g., liblinear-train (the default is svm-train).
Predicting utility, e.g., liblinear-predict (the default is svm-predict).
Temporary directory (the default is /tmp).