. Author manuscript; available in PMC: 2013 Jul 1.

Published in final edited form as: Comput Stat Data Anal. 2012 Jul 1;56(7):2317–2333. doi: 10.1016/j.csda.2012.01.012

Algorithm 1.

Joint Adaptive Mean-Variance Regularization

1. for l = 1 to C_max do

• Select a variable cluster configuration

C

with l clusters.

• Standardize each variable individually using corresponding estimates

{\hat{μ} (l_{j}), {\hat{σ}}^{2} (l_{j})}

where

l_{j} \in {C_{l}}_{l = 1}^{l}

• Compute the corresponding Similarity Statistic estimates

{{\hat{S i m}}_{p} (l)}_{l = 1}^{l}

as in 8.

2. Find the optimal cluster configuration

C

with Ĉ clusters, where Ĉ is determined as in 9.

3. Standardize all variables individually using this optimal cluster configuration

C

. After which, all means

{\hat{μ}}_{j}^{*}

and variances

{\hat{σ}}_{j}^{* 2}

of the transformed data are assumed to follow sampling distributions with target first moments, i.e. (0, 1) respectively.