IoT for Development: Building a Classification Algorithm to Help Beekeepers Detect Honeybee Health Problems Early

Antonio Rafael Braga, Appalachian State University
Edgar E. Hassler, Appalachian State University
Danielo G. Gomes, Universidade Federal do Ceará
Breno M. Freitas, Universidade Federal do Ceará
Joseph Cazier, Appalachian State University

Abstract

Bees are the main pollinators of most wild plant species and are essential for the maintenance of plant ecosystems and for food production. However, in recent years they are suffering from deforestation and pesticides. Here, we propose a method to identify the health status of bee colonies. We trained, validated and tested 4 classification algorithms (Naive Bayes, k-NN, Random Forest and Neural Networks) on actual data from a beehive that was monitored for 6 months. For the generation of the classification model, we take into account data from internal sensors to the hive (temperature, relative humidity, and weight), external data (temperature, pressure, wind speed, and rainfall). We also use data from inspections performed weekly by a specialist in beekeeping. We compared the four algorithms and arrived at a high precision classification model to automatically identify the health status of bee colonies.

 

IoT for Development: Building a Classification Algorithm to Help Beekeepers Detect Honeybee Health Problems Early

Bees are the main pollinators of most wild plant species and are essential for the maintenance of plant ecosystems and for food production. However, in recent years they are suffering from deforestation and pesticides. Here, we propose a method to identify the health status of bee colonies. We trained, validated and tested 4 classification algorithms (Naive Bayes, k-NN, Random Forest and Neural Networks) on actual data from a beehive that was monitored for 6 months. For the generation of the classification model, we take into account data from internal sensors to the hive (temperature, relative humidity, and weight), external data (temperature, pressure, wind speed, and rainfall). We also use data from inspections performed weekly by a specialist in beekeeping. We compared the four algorithms and arrived at a high precision classification model to automatically identify the health status of bee colonies.