Time Series Cross-Validation Techniques for determining the order of the Autoregressive models
P. Kiran Kumar, B. Sarojamma · International Journal of Advanced Research in Computer Science · 2017
In the present research paper, we introduced some techniques of cross-validation for time series data. Six different types of time series cross-validation techniques are presented and also discussed various problems in selecting the initial training sample size and the size of training folds. In this paper, all cross-validation techniques, most appropriate techniques for model selection in time series analysis and advantages of the each technique are discussed with empirical study. Keywords: Time series cross-validation, Prediction error, Order of AR model, Training sample, Test sample.