Energy Stored in Above-Ground Biomass Fractions and Model Trees of the Main Coniferous Woody Plants

The paper considers energy stored in above-ground biomass fractions and in model trees of the main coniferous woody plants (Picea abies (L.) H. Karst., Abies alba Mill., Pinus sylvestris (L.), Larix decidua Mill.), sampled in 22 forest stands selected in different parts of Slovakia. A total of 43 trees were felled, of which there were 12 spruces, 11 firs, 10 pines, and 10 larches. Gross and net calorific values were determined in samples of wood, bark, small-wood, twigs, and needles. Our results show that these values significantly depend on the tree species, biomass fractions, and sampling point on the tree. The energy stored in the model trees calculated on the basis of volume production taken from yield tables increases as follows: spruce < fir < pine < larch. Combustion of tree biomass releases an aliquot amount of a greenhouse gas—CO2, as well as an important plant nutrient, nitrogen—into the atmosphere. The obtained data must be taken into account in the case of the economic utilization of energy stored in the fractions of above-ground tree biomass and in whole trees. The achieved data can be used to assess forest ecosystems in terms of the flow of solar energy, its accumulation in the various components of tree biomass, and the risk of biomass combustion in relation to the release of greenhouse gases.

Meta-Tree Random Forest: Probabilistic Data-Generative Model and Bayes Optimal Prediction

This paper deals with a prediction problem of a new targeting variable corresponding to a new explanatory variable given a training dataset. To predict the targeting variable, we consider a model tree, which is used to represent a conditional probabilistic structure of a targeting variable given an explanatory variable, and discuss statistical optimality for prediction based on the Bayes decision theory. The optimal prediction based on the Bayes decision theory is given by weighting all the model trees in the model tree candidate set, where the model tree candidate set is a set of model trees in which the true model tree is assumed to be included. Because the number of all the model trees in the model tree candidate set increases exponentially according to the maximum depth of model trees, the computational complexity of weighting them increases exponentially according to the maximum depth of model trees. To solve this issue, we introduce a notion of meta-tree and propose an algorithm called MTRF (Meta-Tree Random Forest) by using multiple meta-trees. Theoretical and experimental analyses of the MTRF show the superiority of the MTRF to previous decision tree-based algorithms.

Predicting Students' Attitudes Toward Collaboration: Evidence From Structural Equation Model Trees and Forests

Numerous studies have shed some light on the importance of associated factors of collaborative attitudes. However, most previous studies aimed to explore the influence of these factors in isolation. With the strategy of data-driven decision making, the current study applied two data mining methods to elucidate the most significant factors of students' attitudes toward collaboration and group students to draw a concise model, which is beneficial for educators to focus on key factors and make effective interventions at a lower cost. Structural equation model trees (SEM trees) and structural equation model forests (SEM forests) were applied to the Program for International Student Assessment 2015 dataset (a total of 9,769 15-year-old students from China). By establishing the most important predictors and the splitting rules, these methods constructed multigroup common factor models of collaborative attitudes. The SEM trees showed that home educational resources (split by “above-average or not”), home possessions (split by “disadvantaged or not”), mother's education (split by “below high school or not”), and gender (split by “male or female”) were the most important predictors among the demographic variables, drawing a 5-group model. Among all the predictors, achievement motivation (split by “above-average or not”) and sense of belonging at school (split by “above-average or not” and “disadvantaged or not”) were the most important, drawing a 6-group model. The SEM forest findings proved the relative importance of these variables. This paper discusses various interpretations of these results and their implications for educators to formulate corresponding interventions. Methodologically, this research provides a data mining approach to discover important information from large-scale educational data, which might be a complementary approach to enhance data-driven decision making in education.

Identificação de Sistemas Dinâmicos Não Lineares Utilizando Modelos Neuro-Fuzzy Lineares Locais com um Algoritmo LOLIMOT-PSO

Este trabalho propõe o uso do algoritmo de otimização baseado em enxame de partículas para a determinação dos pontos de divisão do subespaço de uma dada dimensão de entrada utilizando algoritmo de treinamento de modelos Neuro-Fuzzy conhecido como LOLIMOT (Local Linear Model Trees). A proposta foi avaliada em dois sistemas dinâmicos não lineares, sendo um modelo NARX (Nonlinear Autoregressive Exogenous) e um sistema de nível. Simulações de Monte Carlo foram efetuadas para analisar o efeito da inicialização aleatória do algoritmo PSO. Os resultados foram comparados com o algoritmo LOLIMOT convencional e em todos os casos foi possível observar uma melhora com relação a função de custo.