Combining expdata with the same xdata about qiskit-experiments HOT 5 CLOSED

For the plot part, what about just not connecting the points and making a scatter plot instead?

from qiskit-experiments.

In #62, I've removed the connecting line, as @coruscating suggested. It still doesn't look very nice because of pairs of dots on the same vertical line, but at least the straight line has disappeared. I vote for merging even before the fit. There is however one non-trivial point, which is, how close do two x's have to be in order to be considered equal.

from qiskit-experiments.

I hesitate to process data like that—what if there's information in the spread of y's for the same x that we lose from merging? I think plotting with merged values and std error bars could be a good compromise. As for how close...maybe np.isclose for now? Though the tolerance has to be adjusted for very small values.

from qiskit-experiments.

I prefer the merge as the default solution, but we could provide an option so the user can create a scatter instead.

from qiskit-experiments.

Will leave as-is like a scatter and wait for experimentalists's opinion.

from qiskit-experiments.