Research and projectsData analysis tools pLDA

Penalised Reduction & Classification Toolbox

Author: Eva Janoušová
Institute of Biostatistics and Analyses, Faculty of Medicine and the Faculty of Science , Masaryk University, Brno, Czech Republic

Citation: Janoušová, E.. Penalised Reduction & Classification Toolbox, version 1.0 [online]. 2016 . Available from:

Penalised Reduction & Classification Toolbox provides algorithms for reduction and classification of various types of data, such as genetic data, two-dimensional (2-D) face image data or three-dimensional (3-D) brain image data. The algorithms were implemented as functions in MATLAB® environment. Nowadays, the toolbox enables reduction of data by selecting most discriminative features using penalised linear discriminant analysis (pLDA) with resampling, penalised linear regression (pLR) with resampling, and t-test or feature extraction using intersubject principal component analysis (isPCA). The reduced data are then classified into two groups using linear discriminant analysis (LDA) or linear support vector machines (SVM). Classification performance of methods acquired by leave-one-out cross-validation can be compared using the McNemar’s test.

documentation (PDF file, 442 kB)
source code (ZIP archive, 337 kB)

Acknowledgement: The toolbox was developed within the frame of the grant project Advanced Methods for Recognition of MR brain images for Computer Aided Diagnosis of Neuropsychiatric Disorders, supported by the Internal Grant Agency of the Czech Ministry of Health (project no. NT 13359-4).