Search

The present study compares the accuracy of automatic univariate time series forecasting methods for the value of the National Consumer Price Index of Mexico, for which monthly data from 1969 to 2021 are used. Two of the main methods are presented in the study area: 1) The Exponential Smoothing State Spatial Models (ETS) and 2) The Autoregressive Integrated and Moving Average (ARIMA) models. A computational algorithm is developed for the identification of a data generating process to estimate the optimal parameters. The second method presents a better performance in the calculation of the predicted values. The results show the usefulness of automatic forecasting methods when proper procedures are followed. [ABSTRACT FROM AUTHOR]

The objective of this analysis is to find the best hierarchical model to forecast the total demand for regular gasoline in Bogotá, Colombia and, therefore, the collection of gasoline surcharges, which is an important tax used to finance road networks and massive transportation systems. We used data reported by 6 wholesalers of regular gasoline in the city, and used two univariate approaches (ARIMA and exponential smoothing (ETS)), five methods and different minimization algorithms to forecast gallons of regular gasoline. Results show that the best combination of these parameters is an ETS model under a simple univariate forecast. [ABSTRACT FROM AUTHOR]

The objective of this work is to predict the consumption of residential electrical energy in the Cajamarca Region through HoltWinters models. This predictive modeling procedure is useful for the short and medium term prediction of electricity sales. The forecast of electricity consumption is important in regional and national energy planning; from their results, the agents of the electric power market make more adequate decisions for their work. The Holt-Winters method was estimated for different smoothing constants, which includes methods for seasonal additive and multiplicative patterns. The value of the statistical tests of the mean absolute percentage error (MAPE) was considered as the main estimator of the capacity of the model. The additive Holt-Winters model was selected as the best prediction model, with an average absolute percentage error of 1,70 and with estimated smoothing constants of 0,4. [ABSTRACT FROM AUTHOR]

The combined application of two techniques improved for zone tension smoothing: Superconvergent Patch Recovery (SPR) and Recovery by Equilibrium Patches (REP), are analyzed. The techniques are improved on hexahedrons for three-dimensional domain problems which allow obtaining more exact results when using the finite elements method. To analyze the combination behavior of these techniques, an h-adaptive procedure is applied using the Zienkiewicz-Zhu estimator to determine the error that involves the smoothed stress field. The convergence results and efficiency index show that both techniques are analogous when the meshes have high degrees of freedom. [ABSTRACT FROM AUTHOR]