Hello, I am using DL4J and want to set up a RNN with LSTM layers. I have my data(which has already been transformed) in a CSV File. Here is an example of what a row of my data looks like:

45.414001,10358500,45.698002,44.728001,0.0

The first row is the value I want to predict or my feature, and the last row is 0.0 because it is a place holder, it will have normal numbers in the future. In the file there are 1260 rows of data just like these, and I want to use them to do a time series prediction using DL4J, predicting one time step into the future. Can someone please walk me through what the next steps are in order to get this data into an object that can be used as input for my Network? Thank you in advance!

thats a super broad question. check out my github where I do a lstm regression with stock data

Ive been reading your example. Could you explain what you did here: ` normalization.preProcess(test);

```
network.rnnClearPreviousState();
INDArray out = Nd4j.zeros(1);
ArrayList<INDArray> output = new ArrayList<>();
out = network.output(test.getFeatures());
output.add(out);
normalization.revertLabels(out);`
```

This is lines 166 to 178 in your dual LSTM stock one. Thank you for providing your examples, they are very helpful.

It looks like I’m using my normalizer to prepare the data and then revert the networks output be the same amount. The output arraylist looks like i didn’t need it after all so I just take the out array and use it in my chart

When i implement your code, it gives me a big array of what i assume to be the outputs for each time step. How would i change that to give me the next time step only. Say I have 150 days of google stock data, and I want the prediction for the 151st day. Do you know how to do that? Thanks.

Try the recurrent regression project. Its simpler and I think will get you more familiar with getting the data into it.