| Type: | Package |
| Title: | Time Delay Spatio Temporal Neural Network |
| Version: | 0.1.0 |
| Depends: | R (≥ 4.2.3), nnet |
| Description: | STARMA (Space-Time Autoregressive Moving Average) models are commonly utilized in modeling and forecasting spatiotemporal time series data. However, the intricate nonlinear dynamics observed in many space-time rainfall patterns often exceed the capabilities of conventional STARMA models. This R package enables the fitting of Time Delay Spatio-Temporal Neural Networks, which are adept at handling such complex nonlinear dynamics efficiently. For detailed methodology, please refer to Saha et al. (2020) <doi:10.1007/s00704-020-03374-2>. |
| Encoding: | UTF-8 |
| License: | GPL-3 |
| NeedsCompilation: | no |
| Packaged: | 2024-05-25 12:00:30 UTC; pc |
| Author: | Mrinmoy Ray [aut, cre], Rajeev Ranjan Kumar [aut, ctb], Kanchan Sinha [aut, ctb], K. N. Singh [aut, ctb] |
| Maintainer: | Mrinmoy Ray <mrinmoy4848@gmail.com> |
| Repository: | CRAN |
| Date/Publication: | 2024-05-26 17:10:02 UTC |
Spatio-Temporal Neural Network
Description
The STNN function fits a Time-Delay Spatio-Temporal Neural Network model for space-time time series data.
Usage
STNN(data,lag, weight0, weight1,hs, h)
Arguments
data |
Spatio-temporal time series (ts) data. |
lag |
Number of time lag(s). |
weight0 |
Zero-order weight matrix. |
weight1 |
First-order weight matrix. |
hs |
Number of hidden layer(s). |
h |
The forecast horizon. |
Details
This function enables you to apply the Time-delay Spatio-Temporal model for analyzing space-time series data.
Value
Model Summary |
Summary of the fitted STNN |
Fitted values |
Fitted values of STNN |
Forecasted values |
h step ahead forecasted values employing STNN |
Author(s)
Mrinmoy Ray, Rajeev Ranjan Kumar, Kanchan Sinha, K. N. Singh
References
Saha, A., Singh, K.N., Ray, M. et al. A hybrid spatio-temporal modelling: an application to space-time rainfall forecasting. Theor Appl Climatol 142, 1271–1282 (2020).
See Also
ANN
Examples
ts.sim1 <- 50+arima.sim(list(order = c(1,1,0), ar = 0.7), n = 100)
ts.sim2<-70+arima.sim(list(order = c(1,1,0), ar = 0.8), n = 100)
weight0=diag(1, 2, 2)
weight1=matrix(c(0, 1, 1, 0), nrow= 2, ncol = 2, byrow = TRUE)
zz=as.matrix(cbind(ts.sim1,ts.sim2))
data=zz
lag=1
hs=2
h=5
STNN(data,lag,weight0,weight1,hs,h)