daltoolboxdp (version 1.2.737)

ts_conv1d: Conv1D

Description

Time series forecaster using a 1D convolutional neural network. Wraps a PyTorch implementation via reticulate.

Usage

ts_conv1d(preprocess = NA, input_size = NA, epochs = 10000L)

Value

A ts_conv1d object.

Arguments

preprocess

Optional preprocessing/normalization object.

input_size

Integer. Number of lagged inputs per training example.

epochs

Integer. Maximum number of training epochs.

References

LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-Based Learning Applied to Document Recognition. Bai, S., Kolter, J. Z., & Koltun, V. (2018). An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling.

Examples

if (FALSE) {
# Conv1D forecaster expects features + 't0' target internally; the R wrapper
# builds the required data frame when you call do_fit/do_predict via tspredit.

tsf <- ts_conv1d(input_size = 12, epochs = 1000L)
# model <- daltoolbox::fit(tsf, your_data)
}

# See:
# https://github.com/cefet-rj-dal/daltoolbox/blob/main/timeseries/ts_conv1d.md