Aplicación para el análisis y comparación de serias cronológicas
Archivos
Fecha
2012-07-03
Autores
Rodríguez González, Dorian
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
El pronóstico basado en series cronológicas es un problema común en la economía, la medicina y otras ramas que utilizan la estadística y métodos de minería de datos. Este proyecto propone y desarrolla una solución computacional ante la necesidad de tener, en un sistema como Weka, una herramienta que permita realizar diferentes pruebas para el análisis de series cronológicas. Se incluyen dos métodos de análisis de series temporales: el primero es para hallar el retardo óptimo que correlaciona dos series, y el segundo permite encontrar la ecuación iterada de regresión entre dos series correlacionadas. En adición, el nuevo paquete provee a Weka con la capacidad de mostrar las graficas de las series temporales y los correlogramas, con el objetivo de facilitar la determinación de los parámetros del modelo ARIMA de una serie. La solución propuesta utiliza como recurso computacional al paquete estadístico R para el trabajo en Weka. Este trabajo expone los detalles del diseño e implementación de la herramienta, explica mediante un ejemplo, el procedimiento para hallar la ecuación iterada de la regresión entre las series de dos variables económicas y finalmente brinda un manual para el uso de la aplicación.
The forecast based on time series is a common problem in economics, medicine and other fields that use statistics and data mining methods. This project proposes and develops a computational solution for the need of having in a unique system like Weka, a tool that allows performing various tests for time series analysis. This includes two methods of time series analysis: the first is to find the optimum lag that correlates two series, and the second allows finding the iterated regression equation between two correlated series. In addition, the new package provides the ability to display graphs of time series and the correlograms, in order to facilitate the identification of the ARIMA model parameters of a series. The proposed solution uses the R statistical package as a computational resource for the work developed by Weka. This paper presents the design details and implementation of the tool, explained through an example, the procedure to find the iterated equation of the regression between the two sets of economic variables. And at the end of this paper it is provided a user manual for the correct use of the application.
The forecast based on time series is a common problem in economics, medicine and other fields that use statistics and data mining methods. This project proposes and develops a computational solution for the need of having in a unique system like Weka, a tool that allows performing various tests for time series analysis. This includes two methods of time series analysis: the first is to find the optimum lag that correlates two series, and the second allows finding the iterated regression equation between two correlated series. In addition, the new package provides the ability to display graphs of time series and the correlograms, in order to facilitate the identification of the ARIMA model parameters of a series. The proposed solution uses the R statistical package as a computational resource for the work developed by Weka. This paper presents the design details and implementation of the tool, explained through an example, the procedure to find the iterated equation of the regression between the two sets of economic variables. And at the end of this paper it is provided a user manual for the correct use of the application.
Descripción
Palabras clave
Serias Cronológicas, Análisis, Comparación, Solución Computacional, Series Temporales, Weka, Modelo Arima, Paquete Estadístico R