Produce a forecast based upon multiple time series with variable lag

Question

Firstly, I'm not a data scientist, but I am keen to understand the power of the subject and have invested some time in learning - most examples of time series analysis, however, consider only a single date series.

I'm looking for advice on the most appropriate methods to learn/practice to allow me to develop a signature (prediction) of how one time-lagging variable affects another. The lag of cause/effect of one variable upon the other is variable depending on river flow, but likely to be between 2 and 8 hours.

I would need to firstly demonstrate some evidence of correlation, accounting for the lag which I am able to calculate in any given circumstance, and then a mechanism of forecasting.

The intention is to ultimately have a system automatically respond to a given situation by preparing in advance for an upstream change by reference to a signature, which would otherwise have reacted to the change as it arrived (and thus less effectively).

Any prompts in the right direction appreciated.

Answer 1

You could look into neural nets, e.g. using LSTM layers. There are examples, such as „Jena“ weather forecast with Keras:

• https://www.tensorflow.org/tutorials/structured_data/time_series
• https://keras.io/examples/timeseries/timeseries_weather_forecasting/

Alternatively, you can use VAR models (vector autoregressive models). If you use R, have a look here (section "Multivariate Time Series Models").