Impact of Chaining Method and Level of Completion on Accuracy of Function Structure-Based Market Price Prediction Models
The goal of this paper is to explore how different modeling approaches to construct function structure models and different levels of model completion affect the information contained within the respective models. Specifically, the models are used to predict market prices of products. These predictions are compared based on their accuracy and precision. This work is based on previous studies on understanding how function modeling is done and how topological information from design graphs can be used to predict information with historical training. It was found that forward chaining was the least favorable chaining type irrespective of the level of completion. Backward chaining models work relatively better across all completion percentages, while Nucleation models don’t perform as well for a higher completion percentage. Hence, a greater attention is needed to understand and employ the methods yielding the most accuracy.