IIND-4103 Forecasts and Time Series
A balanced consideration of factors influencing future events and perfect familiarity with the various existing forecasting techniques are required in order to properly make a forecast. The course’s main objective is studying techniques and tools which allow identifying the behavior or pattern of a time series (observations or data taken periodically throughout time) in order to predict its future behavior. The course has three modules: in the first one a review is made of the fundamentals of traditional forecasting methods for trend processes and for processes exhibiting some sort of seasonal pattern, in the second, ARIMA models and their GARCH extensions and transfer models are presented, the third module alternates sessions presenting cases illustrating the construction, identification and validation phases of the model with workshop sessions making extensive use of computers. Prerequisite: Linear Models or solid knowledge of statistics.
Catalog page for this course