Skip to main content

Learners : Matchbox recommender : Command-line runners

Trainer

Training is performed using the Train argument to Learner Recommender. It takes as input a training dataset and outputs a serialized trained model, which can be loaded later for making predictions. Training takes in a number of arguments, explained in the Setting up a recommender section. There is more detail on the training procedure in the Training section.

Required parameters

Optional parameters

Example

Learner Recommender Train --training-data TrainingSet.dat  
                          --trained-model TrainedMatchbox.bin   
                          --traits 5  
                          --iterations 30  
                          --batches 4  
                          --use-user-features   
                          --use-item-features