I'm building my arima model and I have a quarterly data. The total dataset is 104 (1996-2021), now how many train data and test data should I use? Inorder to have an accurate forecast data. Thanks

CodePudding user response：

For ARIMA models, the rule of thumb is that you should have at least 50 but preferably more than 100 observations.

You can split the whole datasets at a ratio of 80 to 20. 80% for training and 20% for testing.