TIME-SERIES ANALYSIS AND FORECASTING
The objective of the course is to provide students with the tools to understand the dynamics of an economic, business or financial time series, for the production of forecasts in the absence of structural changes as well as for the modeling of the relations between two or more phenomena analyzed over time.
Expected learning outcomes
Knowledge and ability to understand
The student has to show knowledge of the main statistical methodologies to describe and interpret the temporal dynamics of one or more economic or financial variables. He/she will also have to demonstrate knowledge of the principles underlying the theory of point and interval forecasts.
Ability to apply knowledge and understanding
The student has to show that he/she is able to apply the procedure of identification, estimation and diagnostic of a statistical model to a real time series and to be able to produce forecasts of a phenomenon both as a number, and for a defined level of confidence. For the estimation of the models he/she will have to know how to use the open source software R.
Autonomy of judgment
The student has to show that he/she is able to choose the most appropriate model for the statistical representation of a specific phenomenon. Furthermore, he/she will be able to interpret the results of the analyses in order to indicate the best strategy under conditions of uncertainty. He/she will also have to show that he is able to exhibit in statistical terms a cognitive requirement in the study of economic-financial phenomena over time.
The student has to show that he/she is able to present the methods, results and statistical interpretation of a study both to specialists in the statistical and economic-financial field and to non-expert people.
The student has to show a good learning of the topics covered and to be able to deepen his knowledge on relevant bibliographic references. He/she will also have the ability to integrate his knowledge by adapting to the evolution of the discipline.
Passing the exam of Statistics.
I part (12 hours). Classical analysis of time-series. Trend and seasonality. Time-series analysis in R.
II part (24 hours). Stochastic analysis of time-series. Global and partial autocorrelation. AutoRegressive (AR), Moving Average (MA) and ARMA models. Seasonal models. Box-Jenkins procedure: identification, estimation and diagnostics. Forecasting theory. Nonstationary time-series. ARIMA models. Seasonal ARIMA models. ADL models. Time-series analysis in R.
III part (12 hours). Multivariate time series. VAR models. Granger causality. Cointegration.Time-series analysis in R.
IV part (24 hours). Financial time-series. ARCH and GARCH models. Asymmetric heteroskedastic models. Volatility forecasting. Value-at-Risk estimation. Time-series analysis in R.
Lectures with applications to real time series using the open source software R.
CRYER CHAN, Time Series analysis with applications in R, Springer
Oral examination. The student has to show to know the main statistical models for the analysis of time series. Moreover, the student has to provide the forecast of a univariate time series using the open source software R, after identifying the most appropriate statistical model.
Lectures are in Italian. The
professor is fluent in English and is available to interact with students in English, also during the examination.