Computing ranks from predictions

Nobody has claimed this yet.

Hello, I'm trying to compute the ranks matrix from the model predictions. using the function below

` mask = ~test_interactions.A.astype(bool)
predictions = model.predict(user_ids, item_ids)
ranks_array = predictions.copy()
argsorts = np.argsort(ranks_array, axis=1)
rankings = range(argsorts.shape[1])[::-1]

for i in range(ranks_array.shape[0]):
ranks_array[i, argsorts[i]] = rankings

ranks_array[mask] = 0

return sparse.csr_matrix(ranks_array)`

When I compare the obtained ranks with the obtained ones from model.predict_rank() I dont get the same ranks.

What am I doing wrong? Can you help please Thank you

Start by reading the model.predict() and model.predict_rank() entry points, then compare their ranking behavior with the supplied NumPy sorting code. Reproduce the mismatch using the shown inputs and determine which rank semantics differ; done means the discrepancy is explained and the expected behavior is documented or covered by a regression test.

Written by the indexing model from the issue text.