• Simple simulation examples:
  • Without signal: simple contrast example
  • With signal: simple contrast example2
• The contrast matrix to subtract the mean:
  • Without signal: Compare with mean without signal
  • With signal: Compare with mean with signal
  • The estimated covariance in the the previous analysis was based on the data excluding the discarded column. Another way is estimating the covariance based on the full data. If we discard the ith column, then discard the ith row and ith column of the estimated covariance matrices. The mashcommonbaseline method is robust of the choice of the discarded column. Compare with mean with signal (Whole)
  • Treat Mean as constant: Mean Constant
  • When we increase R: MeanSignalR=50
  • Compare commonbaseline with Bovy: Compare with Bovy
• Comparison:
  • Compare mash with miximash: Mash vs Miximash
  • Compare mashcommonbaseline with miximash: Mashcommonbaseline vs Miximash

• Outlier There are two kinds of outlier problem in mash.

  • The first one is that the covariance matrices used to fit the mash model are not enough. The correct covariance matrices are not included in the mash model.

  • The other one happened when we want to apply the fitted mash model on a new dataset. The new data contains covariance structures that not currently coverred in the fitted model. Outlier

---

This R Markdown site was created with workflowr

---