Analysis of Biological Data Based on the $k$-Means Method for Development of Detection Systems for Calving Phase of Cows

In this paper, we analyze body surface temperature data of cows for development the detection system for the calving phase. Firstly, we perform preprocess to remove the noise in the body surface temperature. Secondly, approximate waveforms based on Fourier series expansion are generated, and the frequency components which does not correspond to calving phase is removed. Next, by using the feature parameters obtained from the approximate waveform, the k-means method is applied to generate a cluster for the normal phase for each individual of cows. Then, we analyze the temperature characteristic based on the variation of feature parameters, and discuss the effectiveness of the detection system for calving phase.