A Novel Clustering-Based Season Factor Approach for Broiler Breeding

Season factor plays an important part in broiler breeding, but it is hard to measure. This paper presents a new approach to season factor, based on two observations. First, it is easier to find that broiler grows slow when the air temperature is too high. Second, along with the increasing of day age, the weight of broiler increase, but the weight gained seems uncertain, especially for different season. Motivated by these observations, we propose a novel clustering-based season factor approach in broiler breeding. First, we cluster four breeding seasons according to the local ten-day mean air temperature, which can catch the demarcation point of two coterminous seasons. We leverage a clustering algorithm based on the most well-known partitioning method, K-Means algorithm, to the above clustering task. Second, we analyze the seasonal broiler growth curve using trimmed mean. We use the broiler growth dataset of the most famous poultry raising company in China to evaluate our approach and the results show the effectiveness of our approach.