ABOUT A PROMPT STRATEGY FOR ESTIMATING MISSING DATA IN LONG TIME SERIES
Fabio H. Nieto, F. Torres Ruiz · Revista de la Academia Colombiana de Ciencias Exactas Físicas y Naturales · 2024
A quick practical strategy is proposed for estimating missing data in time series that obey low-order ARIMA models and whose length is greater than that supported by current statistical computer programs. The proposed methodology is based on the idea of identifying the series model from its subseries. To obtain these subseries, a minimal number of data points following a missing observation are deduced to achieve numerical stabilization in its recursive prediction.