Class Forecast Mathematical Methods

Software use (Gretl and Eviews) for time series analyses, model estimation/building and forecasting. Skills acquisition researching in national and international databases. Independence in data research, its interpretation and transformations. Identify the main mathematical properties in time series to allow Box-Jenkins methodology application. Perform estimations and forecasting. Critical analysis of the results, comparing the expected with those obtained.

Syllabus 1. Statistics review 2. Linear Regression Model Time series Ordinary Least Squares Restrictions 3. Series components Trend Seasonality Cycles Outliers 4. Stationarity of the series Dickey Fuller Test 5. Box-Jenkins Methodology Autoregressive models Moving Average Models ARMA, ARIMA, SARIMA Models 6. Models estimation and validation 7. Selection of models 8. Forecasting

Valuation: One Intermediate Test (30%), class assessment (Case study with Gretl and time series in the scope Box-Jenkins methodology 10%) and Final Test (50%). Teaching methodology : The course will be driven through practical exercises. Students will learn the several techniques because they will need to apply them in order to solve exercises.

Gujarati, Damodar - Basics Econometrics. 4th, McGraw-Hill Amaro A., 2012,

 Caiado, J. (2016). Métodos de Previsão em Gestão: com aplicações em Excel. 2ª edição. Edições Sílabo Franses, Philip Hans, 1998, Time Series Models for Business and Economic Forecasting, Cambridge University Press.