# Quantum Machine Learning: how to get effective time of training/scoring

I am trying to examine the potential of Quantum Machine learning in terms of performance and time compared to classical algorithms. I am using both Qiskit's QSVM and scikit's SVM with Qiskit Quantum Kernel.

I would like to know if it is possible to get the training and scoring time for these algorithms. A simple time.time() does not take into account the queue on IBM servers and is not useful for the purpose. I know that for circuits one can use job.time_taken, but I don't understand how to use it in this context if it is possible.

I attach the very simple code I am using to define and training the model.

model = SVC(kernel=Qkernel.evaluate, probability=True) #Qkernel is the Quantum Kernel
model.fit(data_train.values, target_train.values)


At https://github.com/Qiskit/qiskit-aqua/issues/628 (Jul 2019) it is claimed that "the job completion for real devices does not currently report executions times". Is that so? How can then be proved the time-quantum advantage with respect to the classical counterpart?