A Common Factor Approach for Multivariate Data Cleaning with an Application to Mars Phoenix Mission Data

• Activity Number: 83
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 4:00 PM to 5:50 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): Dongping Fang* and Wei Ding and Elizabeth Oberlin and Samuel P. Kounaves
• Companies: Zurich and University of Massachusetts and Tufts University and Tufts University
• Keywords: Data Cleaning ; Factor Analysis ; Statistical Learning

Abstract

During the summer of 2008, the Wet Chemistry Laboratory (WCL) on board the Phoenix Lander performed the first comprehensive wet chemical analysis of the soil on Mars. Data collected is very noisy due to unexpected instrumental and environmental factors on Mars.

Previous data analyses have processed the data one ion selective electrodes (ISE) measurement at a time without considering the associations among all ISE signals. This paper proposes a new common-factor removal method that utilizes all ISE measurements simultaneously to find the hidden common factors that drive all measurements to vary simultaneously. These common factors represent the errors and variations caused by the combined and complicated influence of varying temperature, pressure, stirring motion, device malfunction, sensor locations, etc. We have cleaned the data by removing the effects of these common factors. In this paper, we reanalyze the data used in Kounaves et al. paper with our proposed method to show the resulting differences. The contribution of this paper is to provide a new data cleaning method that can be used on all WCL data and can also be used in the future for similar instruments.