Efficient fuzzy frequent pattern tree mining technique to predict chronic kidney diseases

Abstract

Data mining is an important role in huge number of applications, which are manufacturing, aerospace, business organizations, government sectors, and medical industry. In the medical field, the data mining is mostly used for detecting disease and diagnoses. The medical analysis is performed in patients to explore the disease are enormous which results in huge data collection. However, in the large number of records some important data are missing. In a dataset, the presence of missing data is a common problem in statistical analysis and it degrades the performance of the classifier model while using the dataset as a training sample. The Weighted Average Ensemble Learning Imputation (WAELI) was proposed for filling the missing values in the clinical dataset by using single value imputation and multiple value imputation models. Furthermore, the priority-assigning algorithm was assigned along with WAELI to each feature (WAELI-FPA) for selecting the high priority features. In this paper, the performance of WAELI-FPA is further improved by modifying the classifier model such as mining fuzzy frequent pattern tree (MFFPT) and orthogonal partial least square (O-PLS) to achieve the optimal result. Initially, the O-PLS is efficiently performed in the reduction of missing valued after that allocating the priority to each feature in the clinical database. Second, the MFFPT is performed which is initially determined the frequent data and assigning fuzzy membership values then utilizing priority mechanism to predict the kidney disease. The performance of proposed model is tested in terms of accuracy, precision, recall and F-Measure.