Diana Domanska: Forecasting methods

Abstract: Time series analysis and forecasting problems exist in many
fields from finance (e.g. predict price fluctuation in stock market),
economics (e.g. predict demand and supply for commodities), sociology,
politics, forecasts pertaining to natural phenomenons. Importance of an
accurate forecasting model is therefore imperative and different methods
have evolved over the past many years. In this talk, some of these
forecasting methods that are based on data (time series,
multidimensional data) will be discussed. Alternative ways to match
proper method to input data and output phenomenon will also be
discussed. I will then present an overview of a hybrid method, that I
had developed as part of my PhD Thesis, which is a compilation of two
different methods used to forecast natural phenomenon along with a
practical example.