Exploratory analysis between predictive mathematical models, applied to energy production through time series
Main Article Content
Abstract
Introduction: energy at the present time can be considered as an essential element in people's lives, as well as in the development and progress of countries, the energy sector is strategic because it allows the operation and operability of the different sectors. it can be said that energy is indispensable in modern society. Forecasting or inferring what is going to happen in the future allows timely decisions to be made and to anticipate events. This is how it becomes important to know the production of the energy sector in the future. In addition, these predictions can be used as starting elements to generate documents such as medium and long-term energy planning. Objectives: conduct an exploratory study of the best techniques that could assist in the prediction of primary energy production in Ecuador, to evaluate the short-term adjustment efficiency through univariate time series. Methodology: in the research work it was possible to conduct an exploratory study of four predictive models in the energy sector of Ecuador, using two techniques, ARIMA and Holt exponential smoothing, which allowed a reliable approximation of prediction in the production of primary energy in the short term, in three years until 2022, using univariate time series. As for the methodological part used to meet the objectives, it began with obtaining the historical series provided by the Ministry of Renewable Resources and Energy in the technical document called National Energy Balance 2019, the data was processed and outliers were determined using the criterion de Chauvenet, once the database for the analysis was determined, the Box-Jenkins methodology was applied to obtain ARIMA and Holt models. Results: the model that best fits the prediction benefits of those analyzed is ARIMA Model-a (1,1,0) whose expression is: Y_t=3365.526+0.074 Y_(t-1)+ε_t, in addition, it was estimated that Primary energy production for the year 2022 in Ecuador could be 236940.541 kilobarrels of oil equivalent (KBEP), with a fluctuation above and below in the interval of [275511.589 .198369.493](KBEP). Conclusions: based on the data obtained, it can be stated that the predictive models found are strictly autoregressive, that is, they are explicit iterative methods, since they determine the value of Y_t depending on the previous result Y_(t-1), in which they do not intervene the residuals of the errors, this indicates that the component of moving averages does not intervene. The prediction with the first three models a, b, c resulted in an increasing behavior and with model h it remained constant.
Downloads
Article Details
References
Balance Energético Nacional. (2019). Ministerio de recursos renovables y energía. https://www.recursosyenergia.gob.ec/wp-content/uploads/2020/12/Balance-Energetico-Nacional-2019-1.pdf
Barrios, R., Castañeda, M., Pedraza, C., Vásquez, J. D. H., & Ibañez, I. (2016). Eliminación de outliers: una estrategia para reducir la incertidumbre tipo a en la calibración de balanzas. In Simposio de metrología.
del Pino, Mateo. (2009). Previsión de ventas en una gran cadena de tiendas (Master's thesis, Universitat Politécnica de Catalunya).
Espino Timón, C. (2017). Análisis predictivo: técnicas y modelos utilizados y aplicaciones de este-herramientas Open Source que permiten su uso.
García Corona, J. (2020). Una visión didáctica de modelos predictivos para una serie de demanda eléctrica en España.
González Casimiro, M. P. (2009). Análisis de series temporales: Modelos ARIMA.
Gujarati, D., & Porter, D. (2010). Econometría. 673-773, ISBN: 978-607-15-0294-0.
Herrera Granda, D. E. (2019). Predicción de demanda eléctrica mediante la aplicación de modelos ARIMA y SARIMA en lenguaje de programación R–caso de estudio en la Empresa Eléctrica Quito (Bachelor's thesis, Quito, 2019.).
Marqués, M. P. (2015). Minería de datos: a través de ejemplos. Alpha Editorial.
Mauricio, J. A. (2007). Análisis de series temporales. Universidad Complutence de Madrid.
Millán Gordo, E. A. (2019). Pronóstico de la demanda de pasajeros aéreos en nueve aeropuertos regionales colombianos.
Peña, D. M. (2010). Análisis de series temporales. Editorial Alianza.
Pineda, S. E. P., Aguilar, J. A. H., & Arroyo-Figueroa, G. (2017). Aplicación de modelos auto regresivos para la predicción de generación de energía eléctrica a partir de datos eólicos. Res. Comput. Sci., 139, 59-70.
Sánchez Sánchez, D. A. (2018). Modelo ARIMA para el pronóstico de la producción de cacao en el Perú 2012-2018.