Fuzzy clustering methods applied to the evaluation of compost bedded pack barns

This paper concerns the application of fuzzy clustering methods in the evaluation of compost bedded pack (CBP) barns. Fuzzy classifiers are developed to assist decision making regarding the control of variables such as bed moisture, temperature and bed aeration. The idea is to identify interactive factors and therefore promote dairy cattle welfare and improve productivity indices. The data was obtained from 42 CBP barns in the state of Kentucky, US. Details about the data acquisition methodology are given. Well-known clustering methods, namely K-Means (KM), Fuzzy C-Means (FCM), Gustafson-Kessel (GK), and Gath-Geva (GG), are considered for data analysis. The efficiency of the methods is determined by validation indices such as the Xie-Beni criterion, Partition Coefficient, and Partition and Dunn indices. Six classes related to the degree of efficiency of the composting process were identified. The GG method showed to be the most accurate according to the majority of the validation indices, followed by GK. The main reason for the best results is the use of maximum-likelihood-based and Mahalanobis distance measures. Fuzzy modeling results and linguistic information have shown to be useful to help decision making in farms that adopt CBP barns as containment systems for dairy cattle.