How to train Keras LSTM with multiple multivariate time-series data?

问题

I have a mechanical problem as kind of a time series with raw data as follows

        time            dtime   cur         dcur      type  proc    start           end
122088  1554207711521   3140    0.766106    0.130276    0   87556   1554203520000   1554207720000
122089  1554207714411   1800    0.894529    0.089670    0   87556   1554203520000   1554207720000

For every proc, there is a time series with time-instances not exactly in proper intervals. I have data from a set of different procs, each coming from the same type of mechanical problem. The target is to predict the estimated time left in the process from a new random instance of a random process.

So my label is eta = end - time.

I have tried clustering the raw data and use NN regression; and dense NN regression from raw data. But the results are not good enough.

I am thinking of using an LSTM RNN for the time prediction. But I am not sure how exactly should I prepare my data to train the LSTM model. I am guessing I have to create a time-series from each proc. But then I have multiple time-series and I do not know how to handle that.

Length of data samples: 122000

Number of uniques procs: 68 (samples per proc are not equal)

Suggestions are welcome. Thanks .

来源：https://stackoverflow.com/questions/55587697/how-to-train-keras-lstm-with-multiple-multivariate-time-series-data