Skip to main content

Table 1 Contruction of simulation parameters for artificial gene expression data from the multivariate normal distribution

From: Automated multigroup outlier identification in molecular high-throughput data using bagplots and gemplots

Setting Group 1 Group 2 Outliers Parameter values
1 μ=0 d - μ 1=f cJ d f c=0.5,0.75
    μ 2=(f c,−f c)TJ d/2 ρ=0.75
    μ 3=(f c,−f c,f c,−f c)TJ d/4 τ=0.01,0.05
2 μ=0 d κ=5·J d μ 1=f cJ d f c=1.0,1.5
    μ 2=(f c,−f c)TJ d/2 ρ=0.75
    μ 3=(f c,−f c,f c,−f c)TJ d/4 τ=0.05,0.1.,0.2
    κ 1=κ+(f cJ d )  
    κ 2=κ+((f c,−f c)TJ d/2)  
    κ 3=κ+((f c,−f c,f c,−f c)TJ d/4)  
  1. Setting 1 represents data for one group with three outliers. Setting 2 represents data for two group, each with three outliers. The last columns shows the simulation parameters that are varied. The mean vectors in group 1 and group 2 for regular observations are given by μ and κ, and those mean vectors for outlying observations are given by μ 1, μ 2, μ 3 and κ 1, κ 2, κ 3. In this notation, J L denotes a vector of ones of length L and denotes the symbol for the Kronecker product