This tutorial covers how to apply RNN (Recurrent Neural Network) with 2 hidden layers to find best fit and to forecast next data values.

  1. Select data series 'Raw data' on left panel [Points].
  2. To add rectangle area, select menu item <Edit><Areas><Rectangle Area>, and draw area with mouse.
  3. To select area, select menu item <Edit><Select Area>, and click left on rectangle.
  4. Click right and in popup menu select item menu item <Transform red selected points>.
  5. Click button Apply RNN.
  6. In dialog select History points M = 10
  7. Method uses X[i] and M history data values Y[i-M+1],...Y[i] to fit Y[i+1].
  8. Select 10 neurons on first hidden layer and 5 neurons on second hidden layer.
  9. Select 30 data points to forecast.
  10. Click Forecast.
  11. Close the dialog.