Imputation using information fusion technique for sensor generated incomplete data with high missing gap

Deepak Adhikari,Wei Jiang,Jinyu Zhan

DOI: https://doi.org/10.1016/j.micpro.2020.103636

IF: 3.503

2021-01-01

Microprocessors and Microsystems

Abstract:Reliable data is crucial to furnish Cyber Physical Systems (CPS) utilities with quality. Connection error, meteorological extremes, sensing error, human error, etc. are reasons for incomplete data in CPS. Missing data is a leading challenge that confines development of CPS and applications. In this paper, we present Ratio-Based Imputation (RBI) algorithm to handle large missing gap based on combination of data fusion and data mining approach. Imputation is performed by data mining technique whereas analysis is done by data fusion technique. Different simulated datasets were generated from real dataset with random deletion of missing value with range from 6% to 25% where highest missing gap consists of 21% of missing data. Various performance indicators are used to compare accuracy of imputation of missing value. Results demonstrated that proposed method is effective and efficient as results enhanced on average of 26% and 28% based on root mean square error and mean absolute error on highly correlated datasets compared to a popular algorithm. All performance indicators accede to appraise proposed method performs better than existing multiple imputation method we compared. Projected imputation procedure exhibited better accuracy, data structure, and low error under various circumstances.

computer science, theory & methods,engineering, electrical & electronic, hardware & architecture