摘要：The Lee-Carter model and its extensions are the most popular methodsin the eld of forecasting mortality rate. But, in spite of introducing severaldi erent methods in forecasting mortality rate so far, there is no generalmethod applicable to all situations.Singular Spectrum Analysis (SSA) is a relatively new, powerful and nonparametrictime series analysis that its capability in forecasting di erent timeseries has been proven in the various sciences. In this paper, we investigatethe feasibility of using the SSA to construct mortality forecasts. We use theHyndman-Ullah model, which is a new extension of Lee-Carter model, as abenchmark to evaluate the performance of the SSA for mortality forecastsin France data sets.
