How Can You Effectively Explore Uncertainty in Time Series Analysis?

[piper28]

Hello all,

I am new to this forum and was wondering if any of you could help me out.

I have this interview scheduled for next week for which I have to prepare a 10 min presentation on Uncertainty and how to determine it in a time series. Now, there is a wealth of information of the net about uncertainty but nothing concrete when it comes to time series.

I am from an engineering background and have never studied uncertainty in any detail except for errors and measurement bias.
Also, this company is into measurement and data analysis and I am assuming these topics would be focussed towards measurement.

Could someone please point me in the right direction? does it have anything to do with correlation, standard uncertainty and linear regression analysis as I have been reading up on these for last couple of days.

Any help would be much appreciated.

Thanks.

 

[viraltux]

A classical approach to time series are the ARIMA models, there are tons of info about it, for instance you can check http://www.duke.edu/~rnau/411arim.htm or http://people.bath.ac.uk/bm232/EC50162/ARIMA Model.ppt

The uncertainty in ARIMA models is calculated in a simple way too so maybe this will suit your needs for the presentation.

Good Luck!

 

[piper28]

Thank you viraltux for your prompt response. Just wondering, would this approach apply to any time series data acquired from any source?

I think I was just naive enough to assume that only knowledge of the source uncertainties would be required to determine them in time series.

 

[viraltux]

Thank you viraltux for your prompt response. Just wondering, would this approach apply to any time series data acquired from any source?

I think I was just naive enough to assume that only knowledge of the source uncertainties would be required to determine them in time series.

You can apply it to any data regardless the source but, obviously, the ARIMA model becomes really useful if the data have an autoregressive / moving average behavior. Nonetheless, you can always use the variance prediction of the ARIMA model to determine the uncertainty of your time series.

 

[Coder-R]

Piper what is your engineering background. Maybe i could help you.