The Altair Community is migrating to a new platform to provide a better experience for you. In preparation for the migration, the Altair Community is on read-only mode from October 28 - November 6, 2024. Technical support via cases will continue to work as is. For any urgent requests from Students/Faculty members, please submit the form linked here
Stationary data for use in ARIMA model
Learner I
All, I'm trying to build an ARIMA time series model. Web reading tells me my data should be stationary whereas my data set is not (STL decomposition shows trend + strong seasonality).
Q1: does my data need to be stationary for Arima
Q2: How can I make data stationary using RapidMiner?
Thank you,
Bart
Q1: does my data need to be stationary for Arima
Q2: How can I make data stationary using RapidMiner?
Thank you,
Bart
Tagged:
0
Answers
To be exact, you need stationary data for an ARMA model. Thats the reason for the "I" in ARIMA. It stands for integrated and means that first your input data is differentiated d-times (d is a parameter of the ARIMA model) and then an ARMA model is fitted to the differentiated data (the differentiated data is expected to be stationary). The forecast is calculated by integrating the predicted values of the ARMA model. So you could try to use d=1 or 2 in your fitting. Would not recommend to use a higher value, cause this leads to a often unstable fitting (training) of the ARIMA model.
If you have trend and especially if you have strong seasonality, I would even recommend, not to use ARIMA for forecasting. Holt-Winters, Exponential Smoothing and Function and Seasonal Component Forecast all have different ways to handle seasonality.
Hopes this helps
Fabian