The Development and Comparing the Performance of Temporal Fuzzy Neural Network Technique and Temporal Fuzzy Decision Trees Case Study of Suitable Thai Elderly Tourists

Abstract

The purpose of this research is to present a solution to change the value of the temporal fuzzy attribute. A model with data mining technique was developed to solve the problems involved with an effect to the suitability of the decision making of Thai elderly in various tourist destinations in Thailand. This model based on the import factors with the crisp value and fuzzy linguistic term. A temporal fuzzy database system with a design in the Conceptual Meta Schema was applied to collect information in the form of temporal fuzzy attributes. Moreover, this model used a temporal fuzzy attribute matching technique, which consists of the first format of the crisp and fuzzy linguistic term, and the second pattern of the fuzzy linguistic term, and the fuzzy linguistic term expression. The replica models developed between the Temporal Fuzzy Neural Network (TFNN) and the Temporal Fuzzy Decision Tree (TFDT), were compared. The result shown that performance values of the TFNN model was the most valuable, with accuracy value, precision value, recall value and f-measure at 88.9%, 79.0%, 88.9% and 83.7%, respectively. The TFNN model was a structure of 24-3-2, with momentum 0.2, and the learning rate value 0.3. This model provided a form of learning and testing cross-validation folds = 5. Keywords: Temporal Fuzzy Decision Tree (TFDT), Temporal Fuzzy Neural Network (TFNN), Temporal Fuzzy Attribute Matching, Elderly Tourists