![IconGems-Print](images/IconGems-Print.png)
193 – Contributed Poster Presentations: Section on Physical and Engineering Sciences
A Common-Factor Approach for Multivariate Data Cleaning with an Application to Mars Phoenix Mission Data
Dongping Fang
Predictive Analytics, Zurich Insurance
Elizabeth Oberlin
Tufts University
Wei Ding
University of Massachusetts - Boston
Samuel P. Kounaves
Tufts University
Wet Chemistry Laboratory (WCL) on board the Phoenix Lander performed the first comprehensive wet chemical analysis of the Martian soil in 2008 [1 - 4]. The WCL has provided data to estimate concentration of the soluble ions in Martian soil, such as Na+, K+, Ca2+, Mg2+, Cl-, ClO4-, and Li+. The WCL data is very precious, it is the first and the only wet chemistry lab data available so far. Due to unexpected high level of noise, the data cleaning is extremely important. Different data cleaning method may result in significantly different ion concentration estimations. Previous WCL data analyses have processed the data one signal measurement at a time without considering the associations among all signals. This paper proposes a new method that utilizes all signal 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 complicated sources. We clean the data by removing the effects of these common factors. In this paper, we reanalyze the WCL data used in Kounaves et al paper [2] with our proposed common-factor data cleaning method to show the resulting differences. The statistical contribution of this paper is to provide a new data cleaning method.