FcExtremeLearningMachines: Forecasting with Extreme Learning Machines

In Mthrun/TSAT: Time Series Analysis Tools (TSAT)

Forecasting with Extreme Learning Machines

Description

Special case for a multilayer perceptrone feed forward network with backpropagation used for forecasting [Huang et al., 2005]

Usage

FcExtremeLearningMachines(DataVec, Time, SplitAt,

Predictors, ForecastHorizon, No_HiddenLayers = NULL,

Scaled = TRUE, No_TrainingNetworks = 10,

PlotEvaluation = FALSE, PlotIt = FALSE,...)

Arguments

DataVec	[1:n] numerical vector of time series data
Time	[1:n] character vector of Time in the length of data
SplitAt	Scalar 'k' with k<n, index of row where the DataFrame is divided into test and train data
Predictors	[1:n,1:d] data frame or matrix of d predictores with n values each
No_HiddenLayers	Number of hidden layers, see elm
Scaled	Should the data and regressors be scaled before applying the model?
No_TrainingNetworks	Number of Training Networks, see elm
PlotEvaluation	Plot output of training data regarding and network architecture
PlotIt	Simple plot to compare forecasting of future
...	Further arguments passed on to elm

Details

The parameters of hidden nodes (not just the weights connecting inputs to hidden nodes) need not be tuned. These hidden nodes can be randomly assigned and never updated (i.e. they are random projection but with nonlinear transforms), or can be inherited from their ancestors without being changed. In most cases, the output weights of hidden nodes are usually learned in a single step. According to their creators, these models are able to produce good generalization performance and learn thousands of times faster than networks trained using backpropagation [Huang et al., 2005].

Value

Model	Model paramters, see example
Forecast	Forecast, see example
TrainingData	TrainingData, see example
TestData	TestData, see example
Accuracy	ME, RMSE, MAE, MPE, MAPE of training and test dataset in a matrix

Note

For an introduction to neural networks see: Ord K., Fildes R., Kourentzes N. (2017) Principles of Business Forecasting 2e. Wessex Press Publishing Co., Chapter 10.

Author(s)

Michael Thrun

References

[Huang et al., 2005] Huang, Guang-Bin; Zhu, Qin-Yu; Siew, Chee-Kheong: "Extreme learning machine: theory and applications". Neurocomputing, Vol. 70 (1), pp. 489–501, 2005

See Also

elm

Examples

requireNamespace('ggfortify')
x=fortify(datasets::sunspot.month)
x=ggfortify::fortify(datasets::sunspot.month)
#Example for a bad forecast
results=FcExtremeLearningMachines(DataVector = x$Data,Time = x$Index, SplitAt=2800)

Mthrun/TSAT documentation built on Feb. 5, 2024, 11:15 p.m.