LSTM forecasting time series

Question

diana haron 2020-9-27

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/600538-lstm-forecasting-time-series

评论： ghada alabaidy 2021-5-28

Hi,

I'm currently learning how to use LSTM using the chicken pox example,

link: https://www.mathworks.com/help/deeplearning/ug/time-series-forecasting-using-deep-learning.html

The data in this example is just 1 row with multiple columns. Does anyone know how I can use it with more data sets (multple row and mutiple columns).

It would be appreciate to provide the example or explanation about it.

Thank you

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

ghada alabaidy 2021-5-28

Excuse me, can I know how the test is calculated? I collected data from the accelerometer with a smartphone and I want it to work in LSTM Can you help me with many thanks

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Srivardhan Gadila 2020-10-1

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/600538-lstm-forecasting-time-series#answer_503332

在 MATLAB Online 中打开

Refer to the documentation of trainNetwork to understand what should be the input and target data format & shape.

Specifically: sequences, Y & net = trainNetwork(sequences,Y,layers,options)

Also the following code might help you:

inputSize = [13 11 1];
nTrainSamples = 50;
filterSize = 5;
numFilters = 20;
numHiddenUnits = 200;
numResponses = 5;
layers = [ ...
    sequenceInputLayer(inputSize,'Name','input')
    flattenLayer('Name','flatten')
    lstmLayer(numHiddenUnits,'Name','lstm','OutputMode','sequence')
    fullyConnectedLayer(numResponses, 'Name','fc')
    regressionLayer('Name','regression')];
lgraph = layerGraph(layers);
analyzeNetwork(layers)
%%
trainData = arrayfun(@(x)rand([inputSize(:)' 1]),1:nTrainSamples,'UniformOutput',false)';
trainLabels = arrayfun(@(x)rand(numResponses,1),1:nTrainSamples,'UniformOutput',false)';
size(trainData)
size(trainLabels)
%%
options = trainingOptions('adam', ...
    'InitialLearnRate',0.005, ...
    'LearnRateSchedule','piecewise',...
    'MaxEpochs',300, ...
    'MiniBatchSize',1024, ...
    'Verbose',1, ...
    'Plots','training-progress');
net = trainNetwork(trainData,trainLabels,lgraph,options);