In-depth processing and analysis of Herd Data

The products developed by the company measure a no of significant parameters on individual animals in the herd. In order to fully maximise the potential of the product and the data generated, in-depth data analysis of the raw data was required to extract key figures and allow the herd owner to monitor the day to day progress of the animal and the herd as a whole. The developed algorithms needed to be compatible with existing software and hardware and allow the company to implement future revisions.

Through an interaction initially via the Rubicon centre in CIT the company approached CAPPA to assist with the problem. A significant volume of data is generated by the automated weighing system. The animals’ weight is measured each time it attends a trough for water. The work undertaken included an automated data cleaning algorithm. This was able to deal with scenarios such as where more than one animal was on the instrument generating values which were too large or where the animal only had three legs on the machine generating values which were too low. In addition to this interpolation was applied to deal with missing data. All data went through a time series smoothing model to evaluate the overall trend of each animal over time. This was then used to calculate the water consumed by each animal per day, predict the animals’ weight and to evaluate the growth rate of the animal at the resolution of a single day.