A frequently encountered problem is the fitting of a data-vector by means of a model function with a number of parameters, whose values are unknown. Minimizing the residual sum of squares delivers the least squares estimates. The problem of their precision is only solved in the linear case. We discuss and illustrate the quality of approximate (though asymptotically exact) confidence statements.
|Place of Publication||Eindhoven|
|Publisher||Technische Hogeschool Eindhoven|
|Number of pages||7|
|Publication status||Published - 1974|