Non-Stationary Order of Vector Autoregression in Significant Ocean Wave Forecasting
Fikka Raudiya, Aniq Atiqi Rohmawati, Didit Adytia
Abstract
Fikka Raudiya, Aniq Atiqi Rohmawati, Didit Adytia
Abstract
This paper studies the implementation of non-stationary multivariate time series model to fit the ocean wave data. A model comprises from a regression term and associate with exogenous variables in a particular time horizon. Because of the trend fluctuation in the data leading to unstable process, differentiated data are used in fitting the model. The approach suggested is applied to the finite order of Vector Autoregression for an improvement in prediction simultaneously of ocean wave by carrying out wind-related information to waves. The proposed model is compared with linear simple autoregressive model. The performance of both forecasting procedures is assessed by RMSE of well-known error measures. The forecast based on the proposed methodology indicated that it can be regarded as a promising method for wave ocean prediction, it outperforms using 4-order Vector Autoregression.
OpenAlex reports 2 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
This paper studies the implementation of non-stationary multivariate time series model to fit the ocean wave data. A model comprises from a regression term and associate with exogenous variables in a particular time horizon. Because of the trend fluctuation in the data leading to unstable process, differentiated data are used in fitting the model. The approach suggested is applied to the finite order of Vector Autoregression for an improvement in prediction simultaneously of ocean wave by carrying out wind-related information to waves. The proposed model is compared with linear simple autoregressive model. The performance of both forecasting procedures is assessed by RMSE of well-known error measures. The forecast based on the proposed methodology indicated that it can be regarded as a promising method for wave ocean prediction, it outperforms using 4-order Vector Autoregression.
Key concepts: Autoregressive model, Vector autoregression, Time series, Series (stratigraphy), Significant wave height, Wind wave, Multivariate statistics, Computer science