Best machine learning model to predict times series from initial conditions

Hi all,

I have some time series generated from a set of initial conditions, to make an example I will call them A, B, C and D conditions.

Each time history has its value of A, B, C and D.

I would like to know which is the best machine learning model with which I can develop a trained framework predicting the corresponding time history behaviour giving the A, B, C and D conditions.

Thank you

Answer 1

[Dinesh Yadav]

Hi Giacomo,

For time series forecasting use ARIMA or NARX algorithm.

For time series prediction use LSTM or RNN algorithm.

Hope it helps.

[Giacomo Notaro]

Thank you very much!

[Giacomo Notaro]

Hi, exploring your solutions I didn't find a model which starts from some values to reconstruct a time series.

They only predict the future steps of an existing time series. I need something which knows only the initial conditions, not the first time steps values. It has to build a sequence from scratch based on the experience of initial conditions and their associated feedback resulting time series.