Smart Poultry Management Platform with Egg Production Forecast Capabilities
Nikolajs Bumanis, Armands Kviesis, Anastasija Tjukova, Irina Arhipova, L. Paura, Gatis Vītols
Abstract
Current demands for poultry products dictate the need for technological advancement in the poultry industry. The commercially available solutions are pretty sophisticated in terms of data gathering, processing and visualisation; however, compared to precision agriculture, not enough attention and development is put towards the capabilities to forecast production rates in the precision poultry sector, especially in the Baltic region. Therefore, based on the previously developed database architecture design for a smart poultry farm with multisensor data and retrospect of the already existing Aihen platform, the module for egg production forecasting was developed. Early development raised a set of challenges, i.e., partly inconsistent and imperfect data, insufficient overall amount of historical data. As a solution to tackle these problems a novel approach for egg production forecasting based on the XGBoost algorithm together with the SHapley Additive exPlanations was developed and incorporated into the Aihen platform as an extension.