Hi I am dealing with a multiple parameter Time series data. I would like to remove the Lag associated with my parameters. I have used the cross-corelation to identify the lag associated with my data. I am aware that we can take the difference in our data to account for the lag, is there another way to account lag. I attaching a picture of the cross correlation analysis of two of my parameters.
Thanks
lag.pdf (2.7 KB)
Hi @adeejo
The first question. and what ? why do you want to remove lag I guess you want to build a model the lag to your response will be important, your independent variables could be autocorrelated, but the response is one issue specially with linear model. You have to differentiate the response in this case.
Hope this hope a little difficult to answer as I do not know your end problem.
Best regards
Alain
Hi you are correct, I am trying to build a model that has 12 parameters and 1 output variable. The data is collected from sensors and therefore some parameters will lag other parameters. Lag removal process is commonly done while building models for wind turbines.
Hey,
Not sure about specific modelling process used for sensor received data, but my understanding says that check for those 12 additional variables contribution in explaining your output variable first. I mean, why would you even want to take this data as Time-Series when you already have powerful side variables.
Now let’s suppose, those 12 variables do not really contributes as much you need, in this case you need to model your data as time-series. The real challenge with Time-Series data is to model the “Lag” and not to remove it. Plus, since you have additional variables, do take them into account while Time-Series modelling as “External Regressors”.
Again, this is as per my basic understanding of the data. It’s highly possible that you’re correct about your statement of lag removal.
Hope it helps.
Peace!
