Updating the polynomial!

The polynomial inbuilt in the Bubble-logger predates the Logger started to report “Sum BPM/L” and hence the “Sum BPM/L” was back then calculated in Excel during the process of building the model. As the model takes into account the surrounding pressure and the temperature of the wort, the model is more than just the polynomial, but the polynomial is essential to get best fit with FG, and as such needs to be evaluated continually.

Hence, to better reflect the situation after the Logger started to post “Sum BPM/L” it is time to re-calculate polynomial on the basis of what the logger actually reports of “Sum BPM/L” and hence not to carry over any unintentionally assumptions from the build-process.

The above picture slider shows 3 pictures of the 78 hydrometer measurements rG vs. “sum BPM/L” from the 17 brews done for the model generation in all, and hence, the “All reported” is the polynomial for all data, and the “Calculated” shows the 28 measurements(6 brews) used for building the model before it was added into the logger software, and lately the “Logger Reported” give the latest 60 hydrometer measurements (11 brews) where the logger reports the “Sum BPM/L”.

The keen reader will notice there is different between the R2 value there give info on how good the polynomial fit the data, see Table 1.

Data setPolynomialR-valueMeasurements,
hydrometer
Brews
Manualy calculated-4E-07x^2+0.0092R2=0,980286
logger reported data-3E-07x^2+0,009R2=0,9826011
All data-4E-07x^2+0.0092R2=0,9777817
Table 1: Showing the 3 different polynomial considered, eg. all data vs. latest data based on Logger reported “Sum BPM/L” Last the polynomial of the initial data set also included.

As the R2 value for the “logger reported Sum BPM/L” is the best one it make sense to even further look at the splitting the data:

Tabel 2: Splitting the calculation of FG based on the 2 polynomials.

So splitting the data and ditching the data before the logger posted “Sum BPM/L” seems to be justified and hence my reason for including a new polynomial, e.g. -3E-07x^2+0,009. During so would hopefully give better precision taking the Geomatic mean falls till 1.6 from the former 2,2 (see below Tabel 3 of old data).

If I make a selection and remove 1 hydrometer measurement (e.g. I only hold one measurement of FG for NEIPA5, hence, argument would be to move it till range/scope finding) and use hence 77 measurements then, the polynomial becomes “-3E-07x^2+0,0089, R2=0.978”. This observation also supports the update of polynomial.

The data can be downloaded here:

Old overview of FG

To keep the historic the old overview is attached here below:

Tabel 3: Old data set of final FG vs. calculated FG.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.