def svm_epoch(model_table, dat_table, label, arr, epoch, step=0.1, mu=0.1): ''' Creates a query to update the SVM model ''' return iutil.bismarck_epoch( model_table, dat_table, 'svm(__PREV_MODEL__, %(arr)s, ' '%(label)s, %(step)s, %(mu)s)' % { 'arr': arr, 'label': label, 'step': step, 'mu': mu }, epoch, label)
def logr_epoch(model_table, dat_table, label, arr, epoch, step=0.1, mu=0.1): ''' Generates a query to train the model This will create a query to insert into the state table. @param model_table: name of table to store models @param dat_table: name of table we iterate over @param label: name of label column @param arr: UDF that returns an double array stored as a string @param epoch: the epoch number (first epoch is 1, the 0th epoch is an empty string) ''' return iutil.bismarck_epoch(model_table, dat_table, 'logr(__PREV_MODEL__, %(arr)s, ' '%(label)s, %(step)s, %(mu)s)' % {'arr':arr, 'label':label, 'step':step, 'mu':mu}, epoch, label)
def svm_epoch(model_table, dat_table, label, arr, epoch, step=0.1, mu=0.1): ''' Creates a query to update the SVM model ''' return iutil.bismarck_epoch(model_table, dat_table, 'svm(__PREV_MODEL__, %(arr)s, ' '%(label)s, %(step)s, %(mu)s)' % {'arr':arr, 'label':label, 'step':step, 'mu':mu}, epoch, label)