impute

impute implements a soft‑thresholded nearest centroid classifier with shrinkage for class prediction and gene selection from microarray gene expression data.

Key Features:

• Nearest centroid classifier with soft‑thresholding: Uses soft‑thresholded class centroids as prototypes to improve prediction on microarray data.
• Shrinkage‑based feature selection: Incorporates shrinkage to identify a subset of genes responsible for class separation.
• Competitive performance and interpretability: Produces centroid‑based, interpretable classifiers with reported performance comparable to or better than published methods.

Scientific Applications:

• Small round blue cell tumors: Applied to classification of SRBCTs using gene expression microarrays.
• Leukemia studies: Used to distinguish leukemia subtypes from microarray gene expression profiles.
• Breast cancer research: Employed to identify genes involved in breast cancer classification from microarray data.

Methodology:

Applies a nearest centroid classifier with soft‑thresholding of class centroids combined with shrinkage for gene selection.