Several performance metrics are currently available to evaluate the performance of Machine Learning (ML) models in classification problems. ML models are usually assessed using a single measure because it facilitates the comparison between several models. However, there is no silver bullet since each performance metric emphasizes a different aspect of the classification. Thus, the choice depends on the particular requirements and characteristics of the problem. An additional problem arises in multi-class classification problems, since most of the well-known metrics are only directly applicable to binary classification problems. In this paper, we propose the General Performance Score (GPS), a methodological approach to build performance metrics for binary and multi-class classification problems. The basic idea behind GPS is to combine a set of individual metrics, penalising low values in any of them. Thus, users can combine several performance metrics that are relevant in the particular problem based on their preferences obtaining a conservative combination. Different GPS-based performance metrics are compared with alternatives in classification problems using real and simulated datasets. The metrics built using the proposed method improve the stability and explainability of the usual performance metrics. Finally, the GPS brings benefits in both new research lines and practical usage, where performance metrics tailored for each particular problem are considered.

In this paper, a method to classify behavioural patterns of cattle on farms is presented. Animals were equipped with low-cost 3-D accelerometers and GPS sensors, embedded in a commercial device attached to the neck. Accelerometer signals were sampled at 10 Hz, and data from each axis was independently processed to extract 108 features in the time and frequency domains. A total of 238 activity patterns, corresponding to four different classes (grazing, ruminating, laying and steady standing), with duration ranging from few seconds to several minutes, were recorded on video and matched to accelerometer raw data to train a random forest machine learning classifier. GPS location was sampled every 5 min, to reduce battery consumption, and analysed via the k-medoids unsupervised machine learning algorithm to track location and spatial scatter of herds. Results indicate good accuracy for classification from accelerometer records, with best accuracy (0.93) for grazing. The complementary application of both methods to monitor activities of interest, such as sustainable pasture consumption in small and mid-size farms, and to detect anomalous events is also explored. Results encourage replicating the experiment in other farms, to consolidate the proposed strategy.

Animal location technologies have evolved considerably in the last 60 years. Nowadays, animal tracking solutions based on global positioning systems (GPS) are commercially available. However, existing devices have several constraints, mostly related to wireless data transmission and financial cost, which make impractical the monitorization of all the animals in a herd. The main objective of this work is to develop a low-cost solution to enable the monitorization of a whole herd. An IoT-based system, which requires some animals of the herd being fitted with GPS collars connected to a Sigfox network and the rest with low-cost Bluetooth tags, has been developed. Its performance has been tested in two commercial farms, raising sheep and beef cattle, through the monitorization of 50 females in each case. Several collar/tag ratios, which define the cost per animal of the solution, have been simulated. Results demonstrate that a low collar/tag ratio enable the monitorization of a whole sheep herd. A larger ratio is needed for beef cows because of their grazing behavior. Nevertheless, the optimal ratio depends on the purpose of location data. Large variability has been observed for the number of hourly and daily messages from collars and tags. The system effectiveness for the monitorization of all the animals in a herd has been certainly proved.