Simulation of an Adaptive Model Based on AIC and BIC ARIMA Predictions

Abstract ARIMA model forecasting algorithm is a commonly used time series algorithm, this paper first obtains stable sequence through differential operation, and then from the AR model, as MA even model. Select appropriate for prediction use it adaptive mode design. In field of machine learning, complexity likely to increase, while accuracy improves, models with complex structure usually cause following overfitting problem. order balance reasonably, using indicators AIC (Akaike Information Criterion), well BIC (Bayesian information criterion), make judgments, which achieved by eliciting penalty terms in paper, established (1,1,2) meets requirements.