Computational performance metrics of a code

Question

I made a time series forecast model, but I need to evaluate the computational performance of this model. I used the time (in seconds) as code below but it is not a good measure.

Does anyone know how to make the code to get the number of CPU cycles or is aware of another metric?

start = tm.time()

    for k in range(len(X)):
        prediction = eplkrls.evolve(gamma_bar, alfa_VS, X[k], y[k])
        predictions.append(prediction)

 end = tm.time()

 duracao = end - start