|
Abstract:
|
Collecting data on the same samples from multiple sources has become more common, which often results in data of mixed types such as continuous, binary and zero-inflated. Semiparametric rank-based approach to canonical correlation analysis (CCA) has been shown to have an excellent performance on dealing with mixed types of the data, however, it cannot be applied to data over time. To address this challenge, we propose time-varying CCA with rank-based approach that allows us to study how the associations between the two data sets change over time. For small sample size, we also propose data aggregation based on the change point. We further improve computational efficiency for rank-based approach using polynomial interpolation.
|
