NeuralNet_3L4TS

scalation.modeling.forecasting.neuralforecasting.NeuralNet_3L4TS

The NeuralNet_3L4TS object supports 3-layer regression-like neural networks for Time Series data. Given a response vector y, a predictor matrix x is built that consists of lagged y vectors. y_t = f2 (b dot f(a dot x)) where x = [y_{t-1}, y_{t-2}, ... y_{t-lags}].

Attributes

Graph
Supertypes
class Object
trait Matchable
class Any
Self type

Members list

Value members

Concrete methods

def apply(y: VectorD, lags: Int, h: Int, nz: Int, hparam: HyperParameter, f: AFF, f1: AFF): NeuralNet_3L

Create a NeuralNet_3L object from a response vector. The input/data matrix x is formed from the lagged y vectors as columns in matrix x.

Create a NeuralNet_3L object from a response vector. The input/data matrix x is formed from the lagged y vectors as columns in matrix x.

Value parameters

f

the activation function family for layers 1->2 (input to output)

f1

the activation function family for layers 2->3 (hidden to output)

h

the forecasting horizon (1, 2, ... h)

hparam

the hyper-parameters (use Optimizer.hp for default)

lags

the maximum lag included (inclusive)

nz

the number of nodes in hidden layer (-1 => use default formula)

y

the original un-expanded output/response vector

Attributes