python - Need help understanding code example - LSTM multivariate time series forecasting -
i have dataset of multiple columns , many rows, each column represents different variable , each row different point in time. want predict value of 1 variables @ next point in time, given data of previous time-step. i found this tutorial doing lstm neural net using keras. adapted code data (dropping no columns, no labelencoding since data floats) , ran without error, still don't understand column being predicted. in example, column predicted first column in spreadsheet. tried move column want predict first in dataset, , scatterplots (predicted vs actual y-value) before , after moving reveals had effect. but, still don't understand why - in code column predicted specified? wiki