This dataset contains hours per hours the activity level of cows on four farms. These data were used in Wagner et al. 2021. The dataset contains 4 files, one for each farm. Files are structured as follows:

cow working id; date; hour (integer); the time (s) spent 'walking' during one hour (the cow is positioned 'in alleys'); the time (s) spent 'resting’ during one hour (the cow is in a resting area (typically cubicles)); the time (s) spent 'eating' during one hour (the cow is positioned at the feeding table); the activity level (unitless), which is the weighted sum of the time spent in each activity (with the following weights: -0.23 for resting, +0.16 for in alleys, and +0.42 for eating). Due to the weights, the hourly activity level can range from -828 (i.e. -0.233600) to 1512 (i.e. 0.423600); finally, for each of the 11 types of events, a boolean is provided, for the question "is there this type of event on this hour ?" (i.e. 1 means this type of event were reported for this hour; 0 means that this type of event were not reported for this hour). Note that in fact daily events are reported here at hourly scale.

There are 11 types of events:

oestrus; calving; lameness includes all types of lameness and issues on claw or leg; mastitis includes all types of mastitis; Event LPS is specific to experimentation as it is for administered lipopolysaccharide (LPS) in the mammary gland on one day to induce inflammation; Event acidosis stands for subacute ruminal acidosis; Event labelled 'other_disease’, which contains all other diseases such as colic, diarrhea, ketosis, milk fever or other infectious disease; Event accidents contains all types of accidents such as retained placenta or vaginal laceration; Event disturbance is mild intervention on animal (e.g. late feeding, alarm test, animal tied for injection, claw trimming, drying of the cow) and other issues on the day but that did not concerned management changes; Event mixing is for when cows were mixed or moved to another park. Event labelled ‘management_changes’ contains marginal management such as ration changes or bed filling. This event is reported in the dataset, but is not considered as influent on the animal behaviour;

A final Boolean sum up the information on whether this hour is considered as normal (i.e. if all the booleans (without considering the management changes one) are equal to 0 then this hour can be considered as normal, then the boolean is set to 1).

Dataset 2 was also used in Wagner et al., 2020 https://doi.org/10.1016/j.compag.2020.105233 and in Wagner et al., 2020, https://doi.org/10.1007/978-3-030-59491-6_32