def mog_cross_validate(x):
import rpy2.robjects as robjects
from rpy2.robjects import r as R
from rpy2.robjects.packages import importr
R.library('mclust')
log_predictions = []
for training, validation in k_fold_cross_validation(x, options.x_validate_groups):
r_x = robjects.FloatVector(inv_sigmoid(training))
clustering = R.Mclust(r_x)
predictions = R.dens(
modelName = clustering.rx2('modelName'),
data = robjects.FloatVector(validation),
parameters = clustering.rx2('parameters')
)
log_predictions.extend(np.log(predictions))
return log_predictions
def kde_cross_validate(x, adjust=0.05):
import rpy2.robjects as robjects
from rpy2.robjects import r as R
from rpy2.robjects.packages import importr
log_predictions = []
for training, validation in k_fold_cross_validation(x, options.x_validate_groups):
r_x = robjects.FloatVector(training)
density_args = {
'from' : 0,
'to' : 1,
'adjust' : adjust,
}
kde = R['density'](r_x, **density_args)
predictions = R['approx'](kde, xout=robjects.FloatVector(validation))
log_predictions.extend(np.log(predictions[1]))
return log_predictions
def start_cross_validation(task_ids, T):
for fold, (training, validation) in enumerate(k_fold_cross_validation(T, options.x_validate_groups)):
train_mixtures(task_ids, fold, training, validation)