MilkHub Yield Performance

What do we want yield for?

The key requirement in measuring yield is to properly represent the productivity of an individual cow relative to the rest of the herd. This information can be used to identify best performers for breeding, and worst performers for culling or drying off.

I’ll just do a herd test!

It might be expected that a herd test would be sufficient for this requirement. Unfortunately a single herd test can give a fairly poor representation of the productivity of an individual cow. This is due to the combined effect of many atypical factors including:

• Day to day cow variation
• “Bad hair day” exceptions such as cup slip, plant fault, human error and nil entry
• Short term animal factors such as sore feet or cycling
• Poor letdown due to the disruptive herd test environment
• Milk meter fault or inaccuracy

What is the best representation of an individual cow yield?

The best representation of yield is obtained when the atypical factors above are eliminated.

An ideal solution would be to take many unobtrusive herd tests that would not unduly disturb the cows. The multiple results could be treated to eliminate exceptions and remaining values averaged to smooth out other variations.

This is the principle behind the MilkHub system. It automatically measures yield for every cow every milking in the normal milking environment. Software is used to remove exceptions and smooth the variations. Further, the yield is also normalised by the average for each milking to remove whole herd and seasonal variation.

How good is the MilkHub

The following graphs show the MilkHub performance in comparison to a single herd test.
In this example a 3 consecutive herd test average is used to approximate the representative productivity.

If the results were ideal all the points would lie on a straight line where the MilkHub or single herd test gives the same result as the representative value. The spread of points around this line is an indication of performance. A secondary contribution to spread due to error in the representative value since it is based on only a 3 herd test average.

The top graph shows performance of the MilkHub. The bottom graph shows the performance of a random single herd test in black and the herd test that diverges the most form the representative value in red.

Overall it can be seen that the MilkHub performs better than a single herd test. This is particularly true with low yield measurements where an atypical low reading is most misleading.