|
|
 |
| |
|
|
|
Ambiguity Modelling for Data Science: Uncertainty-Aware Fuzzy Systems and Statistical Validation |
|
|
|
PP: 805-818 |
|
|
doi:10.18576/jsap/140621
|
|
|
|
Author(s) |
|
|
|
Suleiman Mohammad,
Yogeesh N.,
Latha B. V.,
Asokan Vasudevan,
Thirumalesha Babu T. R.,
Mohammad Faleh Hunitie,
|
|
|
|
Abstract |
|
|
| In this paper we show how advanced fuzzy systems can be used in the domain of data science, discussing both
their mathematical underpinnings and applications in practice. The fuzzy systems constitute an architecture in which is possible to deal with and model uncertainty, through a way of making decisions in complex and dynamic scenarios. The paper starts by introducing the basics of Fuzzy Set Theory, including the concept of Fuzzy Sets and its fundamental elements such as Membership Functions and Fuzzy Logic Operations. It continues to describe a number of case studies about fuzzy clustering, regression model, decision tree, anomaly detection and predictive analytics demonstrating how fuzzy logic can be incorporated into imprecise and heterogeneous data. Fuzzy regression application for predictive analytics and anomaly detection in manufacturing process case studies are shown to enhance the precision of prediction as well to improve the operational efficiency of the process. You can combine fuzzy systems with other traditional statistical methods to enhance model predictions that can impact your data science conclusions in many areas. We conclude by setting forth future research directions, stressing on the advancement in the design of membership functions; incorporation with deep learning; extension to different issues such as real-time decision making; and improvement in the interpretability. This study highlights the importance of advanced fuzzy systems to enhance the traditional facets of data issues and revolutionize data-driven decision-
making in its entirety. |
|
|
|
|
|
|
|
