| Popis: |
This chapter discusses the transformation of curvilinear into linear regression. Substantial simplifications in the computation are achieved by the application of the two-point method to the determination of the regression parameters. For the approximating curve to describe properly the distribution of the points on the diagram, the equation of the curve must be a function with more than two parameters. As in linear transformation, it is necessary to take logarithms that provide five-digit numbers, and as the frequency of the distribution is expressed by numbers with three of four significant digits, the computations connected with the determination of the parameters of the regression curve by the method of least squares are cumbersome and time-consuming. Incomparably simpler calculations are required for the two-point method. In demographic research, exponential curves are most frequently used. A hyperbola is used in the theory of costs. |
