def learnModel(n_users, m_items, regU, regI, regJ,
learningRate, R, features, epochs, numberOfIterations):
"""Learns a model with the BPRMF algorithm, actually just calls mf.
n_users -- The highest internal assigned User ID
m_items -- The highest internal assigned Item ID
regU -- Regularization for the user vector
regI -- Regularization for the positive item
regJ -- Regularization for the negative item
learningRate -- The learning rate
R -- A dict of the form UserID -> (ItemId, Rating)
features -- Number of features of the items and users
epochs -- Number of epochs the model should be learned
numberOfIterations -- Number of iterations in each epoch
Returns:
W -- User Features
H -- Item Features
Based on:
Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and
Lars Schmidt-Thieme. 2009. BPR: Bayesian personalized ranking from
implicit feedback. In Proceedings of the Twenty-Fifth Conference on
Uncertainty in Artificial Intelligence (UAI '09).
AUAI Press, Arlington, Virginia, United States, 452-461.
"""
import mf
return mf.learnModel(n_users, m_items, regU, regI, regJ, learningRate, R,
features, epochs, numberOfIterations, logLoss,
dLogLoss)
def learnMF(r, trainingDict, reg, ler, lossF, dlossF):
W, H = mf.learnModel(
r.getMaxUid(), r.getMaxIid(),
reg, reg, reg, ler, trainingDict,
features, EPOCHS,
r.numberOfTransactions,
lossF, dlossF)
# np.savez_compressed(
# "BPRMFModelFile", W=W, H=H)
return W, H
def learnModel(n_users, m_items, regU, regI, regJ,
learningRate, R, features, epochs, numberOfIterations):
"""Learns a model with the RankMFX algorithm, actually just calls mf.
n_users -- The highest internal assigned User ID
m_items -- The highest internal assigned Item ID
regU -- Regularization for the user vector
regI -- Regularization for the positive item
regJ -- Regularization for the negative item
learningRate -- The learning rate
R -- A dict of the form UserID -> (ItemId, Rating)
features -- Number of features of the items and users
epochs -- Number of epochs the model should be learned
numberOfIterations -- Number of iterations in each epoch
Returns:
W -- User Features
H -- Item Features
"""
import mf
return mf.learnModel(n_users, m_items, regU, regI, regJ, learningRate, R,
features, epochs, numberOfIterations, hingeLoss,
dHingeLoss)
