Integrating Deterministic Approach with Data-driven and Emerging Metaheuristic Computing Algorithm for the Estimation of Uncertain Forest Knowledge

This research considered a critical evaluation and estimation of the forests knowledge (FK) among university students in North Cyprus using two different approaches namely; deterministic and artificial intelligence (AI) based models. For the first approach, the study broadly assessed university students' knowledge about the forest in general context, forest protection, the importance of forest, forest administration, the danger of deforestation, individual and the government has taken the responsibility of forest versus their nationalities. Primary data were collected with the aid of a structured questionnaire in line with the proposed five research questions; all the proposed research questions were adequately answered through Cross-tabulation analysis. The outcome of the result revealed that students' knowledge about forest, importance of forest, individual and government responsibility of taken care of forest has no significant effect on nationality; while on the contrarily, the result indicated that students' knowledge about forest protection, poor forest administration and danger of deforestation has significant effect on nationality. For the second approach employed the estimation of forest knowledge using three artificial intelligence (AI) models (ANN, SVM, and ANFIS) and a classical MLR model. The performance efficiency of the models was evaluated using four statistical measures. The results indicated that SVM-M1(R2=0.8893, MSE=0.0950, and R=0.9430) outperformance all the other models despite other AI based models proved reliable for the estimation of forest knowledge. Subsequently, emerging optimization algorithm called hybridized support vector regression (SVR) Harris-hawks optimization (HHO) (i.e., SVR-HHO) were used to predict the FK was employed to hybridized SVR model in order to improve the prediction accuracy of the model. The result indicated that HHO proved promising with more than 99% prediction accuracy.