By that I mean, there is no measurement cherry picking. No "data cleaning" functions to reduce "noise". The measurement would have to use all observations as t

By that I mean, there is no measurement cherry picking. No "data cleaning" functions to reduce "noise". The measurement would have to use all observations as they were observed by the same exact hardware in the same environment day after day. From this dataset the measurement function can interpolate or extrapolate how ever much it wants, but it must use the same algorithm year after year.