Help with function approximation in Mathemacica...
- To: mathgroup at smc.vnet.net
- Subject: [mg94718] Help with function approximation in Mathemacica...
- From: somdevil <rexusfund at gmal.com>
- Date: Sun, 21 Dec 2008 05:53:41 -0500 (EST)
Hello,
I would like to make function approximation(prediction of time series). I have the following code, which predicts the values of future events based on data. But I can't make the code to work for future values...I mean the code below simply predicts the values within a range. But I want further prediction:
Thanks alot in advance!!!
<< NeuralNetworks`
<< LinearAlgebra`MatrixManipulation`
<< Statistics`ContinuousDistributions`
<< Graphics`MultipleListPlot`
<< exe2.csv;
Dimensions[xedema]
u = TakeColumns[xedema, {2, 4}];
y = TakeColumns[xedema, {1}];
ListPlot[Flatten[TakeColumns[xedema, {1}]]]
ue = u[[Range[10]]];
ye = y[[Range[10]]];
uv = u[[Range[1, 10]]];
yv = y[[Range[1, 10]]];
na = {1};
nb = {1, 1, 1};
nk = {1, 1, 1};
{model1, fitrecord} = NeuralARXFit[ue, ye, {na, nb, nk}, FeedForwardNet, {2},
2];
model1[[1]][{yy[t - 1], u1[t - 1], u2[t - 1], u3[t - 1]}]
NetComparePlot[uv, yv, model1, PredictHorizon -> 5, ShowRange -> { 1, 10}]
exo = uv + 10;
exo1 = yv + 10;
model1[{10}]