Aplicación de modelos arima para la proyección del recaudo del impuesto de industria, comercio, avisos y tableros ICA
Con el fin de pronosticar el comportamiento del recaudo del impuesto de industria, comercio, avisos y tableros (ICA), para el año 2016, se plantea la metodología de modelos ARIMA, por su capacidad de predicción a corto plazo. Se definen cuatro fases: La primera es la identificación, describiendo el...
- Autores:
-
Rosas Gualdron, Adriana Marcela
- Tipo de recurso:
- http://purl.org/coar/version/c_b1a7d7d4d402bcce
- Fecha de publicación:
- 2016
- Institución:
- Universidad Industrial de Santander
- Repositorio:
- Repositorio UIS
- Idioma:
- spa
- OAI Identifier:
- oai:noesis.uis.edu.co:20.500.14071/35510
- Palabra clave:
- Recaudo
Impuesto
Estacionariedad
Estacionalidad
Modelo Arima (Modelo Autoregresivo Integrado De Media Móvil)
Proyección.
In order to predict the behavior of tax collection on industry
commerce
boards and advertising (ICA) for the year 2016
the methodology of ARIMA models arises by their ability to predict short term. In order to achieve the goal there are four different phases: The first is identifying
describing the behavior of the historical collection and assess the possibility of converting the stationary series both seasonal and regular part. The second phase consists of the estimation model
taking into account the criteria AIC
small errors and randomness of residuals. The third is validation: at this stage the 2015 collection whose figures already taken have been caused and as the screening of this year in contrast to ensure the quality of the model. The last phase is the prediction
in which case the collection is expected by 2016
taking a confidence interval of 80%. It should be noted that collection is made up the value to pay more interest on arrears and is expressed in thousands of pesos. The frequency of the time series analyzed bimonthly seasonal variation is a very important characteristic to be accounted for in a time series forecasting model. Traditionally
the role of autocorrelation function is used to identify if the process is stationary and then the partial autocorrelation function is used to detect the model. Once
guaranteed the stationarity of the process provides that the historical model behavior of the series is set in an . To validate if the selected model is appropriate
Box Ljung test
Shapiro Wilk test and integrated periodogram are used. Concluding that the residuals are not correlated
normally distributed and randomly.
- Rights
- License
- Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)