Over Sampling for Time Series Classification

Oversampling of imbalanced univariate time series classification data using integrated ESPO and ADASYN methods. Enhanced Structure Preserving Oversampling (ESPO) is used to generate a large percentage of the synthetic minority samples from univariate labeled time series under the modeling assumption that the predictors are Gaussian. ESPO estimates the covariance structure of the minority-class samples and applies a spectral filer to reduce noise. Adaptive Synthetic (ADASYN) sampling approach is a nearest neighbor interpolation approach which is subsequently applied to the ESPO samples. This code is ported from a 'MATLAB' implementation by Cao et al. and adapted for use with Recurrent Neural Networks implemented in 'TensorFlow'.


Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.


0.0.1 by Lan Wei, 2 years ago


Browse source code at https://github.com/cran/OSTSC

Authors: Matthew Dixon [ctb] , Diego Klabjan [ctb] , Lan Wei [aut, trl, cre]

Documentation:   PDF Manual  

GPL-3 license

Imports fields, MASS, stats, utils, parallel, doParallel, doSNOW, foreach

Suggests knitr, rmarkdown, keras, dummies, rlist, pROC, devtools, knitcitations, testthat, xts

See at CRAN