Statistical modelling / Fitting Wood's curve

Romney et al. (2005) also include terms in the model to account for seasonal effects of month of calving and month of sampling but, for the purpose of this case study, we shall not consider these effects.

First we calculate the logarithmic values in CS6Data2 using Spread → Calculate → Column… In doing so we discover that the first milk yield recorded for cow 830 is zero. Therefore, we need to delete this record and make a note in the Audit worksheet of CS6Data1 (see note).

We need to analyse each cow separately. The best way to do this is to unstack the columns for logmilk, logday and day (Spread → Manipulate → Unstack…) but this time we shall tick the box 'Create Suffixed column names' in the dialog box.

The section of the spreadsheet below displays the first few records for logmilk for the first nine cows. The remainder of the logmilk values are to the right off the screen, followed by values for logday and day also hidden.