Sammendrag
Silverman(1986) tells about a 'recent surge' of interest density estimation. One of the theories discussed in the book is the kernel estimate, presented in Rosenblatt(1956) and Parzen(1962). This project is about using parametric models for the same general purpose. Using normal distributions that are log expanded with orthogonal cosine functions, and the Focussed Information Criterion for sub model selection, we hope to compete with the kernel estimate as a general scheme for density estimation, and mode hunting.