Queenless honeybee acoustic patterns

The data set consists of acoustic patterns of five Carniola honeybee colonies located in Zacatecas, Mexico. The colonies were monitored from March to May 2018. The beehives analyzed were two healthy queenright colonies with a huge population, two healthy colonies with moderate populations, and a queenless colony with a low population. Two queens were removed from two colonies to create the queenless condition. The acoustic pattern of healthy colonies is presented in Folder 1, and the experiment of queens removed in Folder 2. The hypothesis is that the queenless state of a beehive can be identified by comparing the acoustic pattern with the pattern of a healthy colony by using machine learning techniques and feature extraction methodologies. The data is presented with no previous preprocessing step, in the analysis Mel frequency cepstral coefficients were used for feature extraction, and the classification was carried out by using support vector machines, neural networks, and logistic regression, all of them with high classification rate. The data can be analyzed by using singular value decomposition or principal component analysis to understand the cluster, and how they change when the queens were removed from the colonies.