ReguCovar: Generate samples by ESPO and ADASYN.

In lweicdsor/OSTSC: Over Sampling for Time Series Classification

Description

Generate samples by ESPO and ADASYN.

Usage

ReguCovar(cleanData, targetClass, ratio, r, per, k, m, parallel, progBar)

Arguments

cleanData	First column stores the label data. The other columns store the sample data (without NA, NaN values)
targetClass	The class to be oversampled
ratio	The oversampling ratio number (>=1) (default = 1)
r	A scalar ratio specifying which level (towards the boundary) we shall push the synthetic data (in ESPO, default = 1)
per	Ratio of weighting between ESPO and ADASYN (default = 0.8)
k	Number of nearest neighbours in k-NN (for ADASYN) algorithm (default = 5)
m	Seeds from the positive class in m-NN (for ADASYN) algorithm (default = 15)
parallel	Whether to execute in parallel mode (default = TRUE). (Recommended for datasets with over 30,000 records.)
progBar	Whether to include progress bars (default = TRUE). For ESPO approach, the bar charactor is |——–|100%. For ADASYN approach, the bar charactor is |========|100%.

Value

newData

lweicdsor/OSTSC documentation built on May 8, 2019, 1:13 p.m.