Introduction. Scots pine plays a special role in Poland, both in terms of nature and economy. An attempt to use neural networks to determine the most important biometric feature, which enables prediction of survivability, may have high utilitarian values. Therefore, the aim of the study was to build a predictive neural model, designed to determine the most important biometric features of trees (scots pine) growing in a forty-year-old stand, which determine the survival and occupied biosocial position.
Material and methods. In order to realize the aim of the study, the collected empirical data from the first ten years and from the thirty and forty years of age of the trees in an unattended stand were analyzed. The research material comes from a permanent research area located in the Siemianice Experimental Forest District (Forest Experimental Department of Poznań University of Life Sciences). Independent variables – diameter and height of trees (each year), were used to build the model.
Results. Two neural models were generated. The first model for survivability analyses in the year 30 (M30) and the second model for survivability analyses in the year 40 (M40). The structure of models m30 and m40 took the form 18:18-36-36-36-1. The generated models were based on multilayer neural network MLP with two hidden layers.
Conclusions. The conclusion drawn with the aid of a generated neural model with the use of net sensitivity analysis indicated the most important biometric features in the prediction of the survivability of common scots pine trees. These are the diameter in the fourth (d4 – rank 1) and fifth (d5 – rank 2) years of age of the tree.
|MLA||Niedbała, Gniewko, and Robert Korzeniewicz. " Determining the significance of biometric features for predicting the survivability of pine (Pinus sylvestris L.) With the use of neural modelling methods." Nauka Przyr. Technol. 12.3 (2018): 297–307. http://dx.doi.org/10.17306/J.NPT.00261|
|APA||Gniewko Niedbała1, Robert Korzeniewicz2 (2018). Determining the significance of biometric features for predicting the survivability of pine (Pinus sylvestris L.) With the use of neural modelling methods. Nauka Przyr. Technol. 12 (3), 297–307 http://dx.doi.org/10.17306/J.NPT.00261|
|ISO 690||NIEDBAłA, Gniewko, KORZENIEWICZ, Robert. Determining the significance of biometric features for predicting the survivability of pine (Pinus sylvestris L.) With the use of neural modelling methods. Nauka Przyr. Technol., 2018, 12.3: 297–307. http://dx.doi.org/10.17306/J.NPT.00261|
Instytut Inżynierii Biosystemów
Wydział Rolnictwa i Bioinżynierii
Uniwersy-tet Przyrodniczy w Poznaniu
e-mail: gniewko@up poznan.pl