Sustainability in beekeeping depends on identifying the factors affecting honey and beeswax yields (HY and BWY) - key products - and accurately predicting these yields. Therefore, this study aimed to predict HY and BWY using a classification and regression tree (CART), eXtreme Gradient Boosting (XGBoost) and Random Forest (RF) algorithms, and thermal image processing in Apis mellifera. In this study, 13 colonies of 6 different breeds raised in 10-frame Langstroth hives were used. The effects of independent variables were predicted using data mining algorithms and 15 performance metrics for the effectiveness of the algorithms. Colony power (CP), thermal temperatures (Tmin, Tmax, and Tmean), breed, a*, b*, red, green, saturation, and brightness impacted HY and BWY in different algorithms, but not birth year of queen, L, hue and blue. As a result, XGBoost, CART, and RF demonstrated high predictive performance, respectively. Due to their higher predictive performance, XGBoost and CART algorithms could predict HY and BWY using CP, thermal temperatures, and image values. These techniques could be useful for producers to monitor production quickly and non-invasively without threatening colony welfare.