Skip to main content

Table 1 Parameters for the point generation under three models

From: Optimal clustering with missing values

Model Mean vectors Covariance matrices Distributions’ hyperparameters
Fixed means and covariances μ1=0·1d, μ2=0.445·1d Σ1=Σ2=0.23·Id
Gaussian means and fixed covariances \(\mu _{1} \sim \mathrm {N}\left (m_{1},\frac {1}{\nu _{1}}\Sigma _{1}\right)\), \(\mu _{2} \sim \mathrm {N}\left (m_{2},\frac {1}{\nu _{2}}\Sigma _{2}\right)\) Σ1=Σ2=0.28·Id m1=0·1d, m2=0.45·1d,
    ν1=30, ν2=5
Gaussian means and inverse-Wishart covariances \(\mu _{1} \sim \mathrm {N}\left (m_{1},\frac {1}{\nu _{1}}\Sigma _{1}\right)\), \(\mu _{2} \sim \mathrm {N}\left (m_{2},\frac {1}{\nu _{2}}\Sigma _{2}\right)\) Σ1IW(κ1,Ψ1),Σ2IW(κ2,Ψ2) m1=0·1d, m2=0.45·1d,
    ν1=30, ν2=5,
    Ψ1=Ψ2=20.7·Id,
    κ1=κ2=75
  1. N, IW, 1d, and Id denote Gaussian, inverse-Wishart, column vector of all ones with length d, and d×d idendity matrix, respectively
\