Using Multivariate Adaptive Regression Splines for Estimating Honey Yield

Abstract

In this study, a model was developed to estimate the Honey Yield (HY) variable, which is an important production parameter in the beekeeping sector. MARS (Multivariate Adaptive Regression Splines) method was used in the creation of the model. Hygiene (HI), Aggressiveness (AG), Brood Frame (BF) and Total Frame (TF) were among the independent variables used in the analysis. In order to better reflect the non-linear structure of the model, the interactions of the independent variables with each other were also included in the model. The performance of the model was examined using critical evaluation criteria such as R2, Pearson correlation coefficient (PC), root mean square error (RMSE) and Akaike Information Criterion (AIC). For the training set, R2 was calculated as 0.677, PC as 0.823, RMSE as 6.307 and AIC as 544.389. These results show that the model can estimate the HY variable quite well on the training data. However, R2 0.567, PC 0.760, RMSE 7.060 and AIC 197.715 values were obtained in the test set. A decrease in the performance of the model was observed in the test set, indicating that the generalization ability of the model may be limited. According to the model results, one of the most effective factors on HY is the HI variable. In cases where aggressiveness is above 1.25 units, a significant decrease in HY was observed. Similarly, positive and negative effects were observed on HY in the ranges where the TC value varied between 5.17 and 8.83. In addition, the effects of the interactions of AG and TF variables with BF on HY were also observed clearly. The model reveals that low hygiene and medium total frame count are important factors in reaching the highest levels of HY. As a result, while the MARS model exhibits a strong performance in the training set, it is observed that its performance decreases slightly in the test set. This suggests that the model requires further improvement. However, in general, it shows that the MARS model is a usable tool for HY estimation and can contribute to the development of strategies to increase HY in beekeeping.

Keywords

• MARS
• Honey
• Aggressiveness
• Brood

References

1. Ağyar O, Tırınk C, Önder H, Şen U, Piwczynski D, Yavuz E. 2022. Use of Multivariate Adaptive Regression Splines Algorithm to predict body weight from body measurements of Anatolian buffaloes in Türkiye. Animals, 12: 2923.
2. Aksoy A, Erturk YE, Eyduran E, Tariq MM. 2019. Utility of MARS Algorithm for Describing Non-Genetic Factors Affecting Pasture Revenue of Morkaraman Breed and Romanov × Morkaraman F1 Crossbred Sheep under Semi Intensive Conditions, Pakistan Journal of Zoology, 51(1): 235-240.
3. Al-Shourbaji I, Alhameed M, Katrawi A, Jeribi F, Alim S. 2021. A comparative study for predicting burned areas of a forest fire using soft computing techniques. 2nd International Conference on Data Science, Machine Learning and Applications, pp: 249-260. https://doi.org/10.1007/978-981-16-3690-5_22
4. Andonov S, Costa C, Uzunov A, Bergomi P, Lourenço D, Misztal I. 2019. Modeling honey yield, defensive and swarming behaviors of italian honey bees (apis mellifera ligustica) using linear-threshold approaches. BMC Genetics, 20(1): 78. https://doi.org/10.1186/s12863-019-0776-2
5. Arthur CK, Temeng VA, Ziggah YY. 2020. Multivariate Adaptive Regression Splines (MARS) approach to blast-induced ground vibration prediction, International Journal of Mining, Reclamation and Environment, 34(3): 198-222.
6. Carroll MJ, Brown N, Huang E. 2025. E-B-ocimene and brood cannibalism: Interplay between a honey bee larval pheromone and brood regulation in summer dearth colonies. PLoS ONE, 20(2): e0317668. https://doi.org/10.1371/journal.pone.0317668
7. Eyduran E. 2020. ehaGoF: Calculates Goodness of Fit Statistics. R package version 0.1.1.
8. Eyduran E, Akin M, Eyduran SP. 2019. Application of Multivariate Adaptive Regression Splines through R Software, Nobel Academic Publishing, Ankara, Türkiye.

9. Friedman J. 1991. Multivariate adaptive regression splines, Annals of Statistics, 19(1): 1-67.
10. Haddad N, Batainh A, Migdadi O, Saini D, Krishnamurthy V, Parameswaran S, Alhamuri Z. 2015. Next generation sequencing of apis mellifera syriaca identifies genes for varroa resistance and beneficial bee keeping traits. Insect Science, 23(4): 579-590. https://doi.org/10.1111/1744-7917.12205
11. Kuhn M. 2020. caret: Classification and Regression Training. R package version 6.0-86.
12. Milborrow. 2020. Derived from mda:mars by Trevor Hastie and Rob Tibshirani. Uses Alan Miller’s Fortran utilities with Thomas Lumley’s leaps wrapper. 2020.earth: Multivariate Adaptive Regression Splines. R package version 5.3.0.
13. Novković N. 2022. Analysis and prediction of production characteristics and prices of honey production in the vojvodina region. Journal of Agricultural Food and Environmental Sciences, 76(7): 23-27. https://doi.org/10.55302/jafes22767023n
14. Padilha A, Sattler A, Cobuci J, McManus C. 2013. Genetic parameters for five traits in africanized honeybees using bayesian inference. Genetics and Molecular Biology, 36(2): 207-213. https://doi.org/10.1590/s1415-47572013005000016
15. R Core Team 2020. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.
16. Rittschof C, Rubin B, Palmer J. 2019. The transcriptomic signature of low aggression honey bees resembles a response to infection.. https://doi.org/10.21203/rs.2.13415/v3
17. Rocha, H. and Dias, J. (2017). Honey yield forecast using radial basis functions. BMC Genomics, (in-press). https://doi.org/10.1007/978-3-319-72926-8_40
18. Shao Y, Tsai Y. 2018. Electricity sales forecasting using hybrid autoregressive integrated moving average and soft computing approaches in the absence of explanatory variables. Energies, 11(7): 1848. https://doi.org/10.3390/en11071848
19. Xu X, Zhou S, Huang J, Geng F, Zhu X, Abou-Shaara HF. 2025. Influence of hyperthermia treatment on varroa ınfestation, viral ınfections, and honey bee health in beehives. Insects, 16)2): 168. https://doi.org/10.3390/insects16020168
20. Zaborski D, Ali M, Eyduran E, Grzesiak W, Tariq MM, Abbas F, Waheed A, Tirink C. 2019. Prediction of selected reproductive traits of indigenous Harnai sheep under the farm management system via various data mining algorithms. Pakistan Journal of Zoology, 51: 421-431.