0readme2.R

#### calibri 2 ####

#' - the two-panel plot is implemented with facet_grid
#' - the advantage is that the scale is the same for residuals and response
#' - the disadvantage is that I cannot control the breaks and that a secondary y-axis does not work
#' (the residuals are signed, which confuses the exp() function)
#' - I have switched to log base e; this was perhaps not a good move but now it is done
#' - all the fits are performed with CONC as independent variable and PEAK_AREA as dependent
#' - this makes sense to me because we can measure PEAK_AREA several times for the same concentration
#' and the error is therefore in the measurement of PEAK_AREA, not in CONC which does not change
#' - the downside of PEAK_AREA ~ f(CONC) is that the function f has to be inverted
#' - when f is a polynomial of degree >1, the inversion can generate negative but also
#' complex roots. Removing roots with a sizeable imaginary part can give problems.
#' - with more than one root it is not clear whether we will always be choosing the correct one
#' - this is why I allow degree >1 to play around with but not for actual processing