Imbalanced-type Incomplete Data Fuzzy Modeling and Missing Value Imputations

Missing values are a common phenomenon in real-world datasets, which caused by many factors such as errors in data acquisition or storage, equipment failure, or human fault in storage. Incomplete data modeling and missing values imputation have become an increasingly important task. Since the regression relationship between attributes is usually different in different clusters, this paper proposes a method called DS-TS-ALI model to model incomplete data that rely on clusters. The precise regression model between attributes is established for incomplete data in the framework of Takagi-Sugeno (TS) fuzzy model. In the premise parameter identification part, a distance density (DS) algorithm based on a partial distance strategy is proposed given the distribution of data categories is imbalanced. Moreover, a membership reconstruction strategy is proposed on this basis. In the consequence parameter identification part, we propose an alternating iterative (ALI) scheme which treats missing values as variables to identify the parameters of the attribute regression model. The imputation will be completed at the end of the modeling process. Experiments on several datasets are conducted to demonstrate the effectiveness of the proposed method.