weka / MIWrapper / 0.1.1

This is the MIWrapper classifier, as implemented in Weka:

Send your input to the algorithm as a simple Json object, with the following options:

mode: One of train/update/load. Use "train" to train a new classifier, use "update" if you have new instances to train an existing updateable classifier, use "load" to load an existing classifier for classifying test data
trainUrl: The path to the training data that you uploaded to our Data API
testUrl: The path to the test data that you would like to get the labels using this classifier (optional if you would like to use the cross-validation option)
modelUrl: The path that you would like to either save the model that is trained or if you would like to load an already trained and saved model
cv: The number of cross-validation folds that you would like to use
options: A string that contains any options that you would like to specifically set for this classifier (format: param name followed by desired value separated by space: e.g. "-C 5 -t 2"). For options specific to this classifier, please see Wekadocs linked above
classIndex: Specify the index of the class values in the test set (caution: When using a test set, assumes that the class values are at the last index if the classIndex parameter is not specified, so please arrange accordingly)