matteo-ronchetti / aws-lambda-power-tuning-ui

UI to visualize the results of https://github.com/alexcasalboni/aws-lambda-power-tuning
Apache License 2.0
99 stars 18 forks source link

Clarify graph interpretation #9

Open alexcasalboni opened 4 years ago

alexcasalboni commented 4 years ago

Another typical misinterpretation is that the "optimal" power value corresponds to the point where the two curves cross.

Additionally, the current implementation creates a somehow misleading interpolation and many users have assumed that intermediate power values will correspond to that (quadratic?) interpolation. While in many cases you'll find interesting patterns if you test additional power values. This is especially important if you test power values that are far from each other, or not equidistant.

In order to solve the first point, do you think we can update the visualization to make sure the two lines never cross? (I think it's useful to plot them on the same chart, but it should be very clear that two independent Y axis are used)

In order to solve the second problem, could we add an option (like a checkbox) to include ALL the power values in the visualization and only plot points corresponding to the (few) tested values?