AdaptiveNormalizer | R Documentation |
R6 class with our adaptation of the method proposed by Ogasawara et al. for one-pass online time-series adaptive normalization. This method is designed to train neural networks and returns the normalized train and test feature sets. In this proposal, the outlier elimination phase has been omitted.
new()
Create a new AdaptiveNormalizer object.
AdaptiveNormalizer$new(wl = 10, maxmin = T, l = 3, returnPoint = F)
wl
Length of the window with historical data values.
maxmin
If TRUE
standardization and normalization are applied, else only
standardization is applied.
l
Times IQR to anomaly removing. It must be a number greater than 0. By default 3, but other common values could be 1.5 and 6.
returnPoint
If FALSE
normalized then the current window is returned, else, the
normalized current data point is returned.
A new AdaptiveNormalizer object.
normalize()
Normalizes the current data value.
AdaptiveNormalizer$normalize(x)
x
Current data value to be normalized.
If returnPoint = FALSE
then normalized current train and test windows are
returned, else, normalized current data point is returned.
denormalize()
Denormalizes the current data value.
AdaptiveNormalizer$denormalize(y)
y
Current data value to be denormalized.
Denormalized current data point.
E. Ogasawara, L. C. Martinez, D. De Oliveira, G. Zimbrão, G. L. Pappa,M. Mattoso, Adaptive Normalization: A novel data normalization ap-proach for non-stationary time series, in: Proceedings of the International Joint Conference on Neural Networks, 2010, pp. 1-8
normalizer <- AdaptiveNormalizer$new(3) normalizer$normalize(10) normalizer$normalize(15) normalizer$normalize(20) normalizer$normalize(10) normalizer$normalize(30) normalizer$normalize(15) normalizer$denormalize(-0.4858841)
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.