JOURNAL OF LIAONING TECHNICAL UNIVERSITY
(NATURAL SCIENCE EDITION)
LIAONING GONGCHENG JISHU DAXUE XUEBAO (ZIRAN KEXUE BAN)
辽宁工程技术大学学报（自然科学版）
MODELING OF VECTOR AUTOREGRESSIVE INTEGRATED MOVING AVERAGE AND GENERALIZED SPACE TIME AUTOREGRESSIVE INTEGRATED MOVING AVERAGE BASED ON DISTANCE INVERSION MATRIX AND CROSS-CORRELATION NORMALIZATION IN THE CASE OF AIR TEMPERA
Nurul Maulida, Georgina Maria Tinungki, Nurtiti Sunusi
Abstract-
Forecasting predicts future events by analyzing historical data. The VARIMA model is used for multivariate time series forecasting, incorporating multiple variables from previous periods. In West Nusa Tenggara, air temperature significantly influences agricultural production and exhibits spatial variations, leading to differences in model parameters across locations. The GSTARIMA model accounts for both spatial and temporal dependencies, allowing parameters to vary by location. This study compares the VARIMA and GSTARIMA models, utilizing inverse distance weighting and cross-correlation matrices to determine the optimal approach for forecasting air temperature. GSTARIMA requires fewer parameters than VARIMA. The results indicate that the GSTARIMA (3,1,1)1 model with inverse distance weighting is the most effective, achieving the lowest RMSE of 2.821 while meeting feasibility criteria. Forecasts from December 2024 to June 2025 closely align with out-of-sample data, demonstrating high accuracy.
Index Terms- Statistics, VARIMA, GSTARIMA, Temperature Forecasting.