Loading Data From CSV’s

It’s useful to know how to load data from CSV files into neural nets, especially when dealing with time series. There’s an easy way to do that with Deeplearning4j:

public static void main(String[] args) throws  Exception {
    RecordReader recordReader = new CSVRecordReader(0,",");
    recordReader.initialize(new FileSplit(new ClassPathResource("iris.txt").getFile()));
    //reader,label index,number of possible labels
    DataSetIterator iterator = new RecordReaderDataSetIterator(recordReader,4,3);
    //get the dataset using the record reader. The datasetiterator handles vectorization
    DataSet next = iterator.next();
    // Customizing params
    Nd4j.MAX_SLICES_TO_PRINT = 10;
    Nd4j.MAX_ELEMENTS_PER_SLICE = 10;

Then configure the neural network and train it on the dataset. This example shows how.

And here is the CSVRecordReader class.

Chat with us on Gitter