A Matrix Completion-based Multiview Learning Method for Imputing Missing Values in Buoy Monitoring Data

Data missing is a common problem in data-driven earth science research. Take the buoy system as an example. Buoy systems have been deployed to monitor coastal waters, and these systems generate massive amounts of data with high temporal resolution. In practice, however, many of these records are often lost, which adversely impacts further data analysis.

In this paper, we propose an efficient model called MC-MVL that combines the advantages of both matrix completion and multiview learning approaches to fill in the missing values. Fig. 1 presents a flowchart of the proposed model. For each buoy, readings of parameters at consecutive timestamps are stored in a matrix that may be incomplete with missing entries. To estimate the missing entries, we first calculate the Euclidean distance to determine the nearest K buoys. Then, the K+1 (the original buoy and its K nearest buoys) data matrices are arranged to form a data tensor. Subsequently, three matrices are derived from the tensor to represent the different views: the temporal-parameter view, the spatial-parameter view and the spatiotemporal view. Each view is formulated as a matrix. The fixed-point continuation with approximate singular value decomposition (FPCA), the non-convex (Non-Convex) and the inexact augmented Lagrange multipliers (IALM) algorithms are separately utilized to reconstruct these three matrices. Then, a ridge regression-based multiview learning algorithm is used to aggregate the estimates of the three views into the final results.

We used a buoy monitoring dataset from the Zhejiang coastal area to verify the imputation ability of the proposed model. The results confirm that MC-MVL achieves better performance than do 8 baseline approaches. Additionally, MC-MVL overcomes the deficiencies of the single-view and two-view methods when faced with block missing patterns.

Fig. 1. Framework of MC-MVL.

Qin M, Du Z, Zhang F, et al. A matrix completion-based multiview learning method for imputing missing values in buoy monitoring data[J]. Information Sciences, 2019, 487: 18-30.