The application of cluster analysis methods in assessment of daily physical activity of dairy cows milked in the Voluntary Milking System

A great individual variability of dairy cows and the diversity of conditions of their maintenance make it difficult to unequivocally interpret the animals' behaviour and, consequently, to assess their welfare objectively. Thus, technical support to cattle breeders seems increasingly important in this respect at the stage of data collection, the analysis of data and assessment of rearing conditions. Therefore, the aim of the project was to examine the possibility of using cluster analysis to assess physical activity of dairy cows milked in the Voluntary Milking System while taking into account environmental conditions. The research included ten Holstein-Friesian cows in June, September, and December of 2015. The data concerning the cows' physical activity were classified with Ward's method and the Kohonen's networks. In general, during individual months, the distribution of cows' average daily physical activity was similar and its variability small. Nevertheless, over the individual months, the following three groups for this feature were distinguished with the use of the cluster analysis: night-morning, morning before noon, and afternoon-evening. At the same time, it was observed that, in at least a few cases, the division could be associated with such environmental conditions as daytime light length changes, temperature, or relative humidity of the air. Therefore, in our opinion, cluster analysis can be helpful in classifying dairy cows physical activity, thus contribute to an objective assessment of behavioural indicators of their welfare.