Cubic Splines and Histosplines

Abstract

In chapter 2 we provided an explanation for the failure of many methods intended to find time- and age-dependent mortality rates from stage structured population data. It was also suggested that an effective method of mortality estimation would be a more local method than those so far suggested. In chapter 4 we present such a method based on surface fitting with cubic spline functions. The method requires some standard techniques for fitting spline functions to noisy point data, which are outlined briefly in section 3.2. Chapter 4 also requires some extensions of spline theory to cover noisy aggregate data and covariance, which are the subjects of sections 3.3 and 3.4 respectively. Some of the development in section 3.3 is new, and section 3.5 briefly discusses some possible areas of application.