ESTIMATING FREQUENCY FUNCTIONS FROM LIMITED DATA.

The article focuses on methods of frequency function estimations using limited data. It states that the use of sparse data in frequency function estimations can be used in determining the expected value that added data would provide, and if its value exceeds the cost of gathering the data. It comments on the use of constrained maximization of entropy in frequency function estimation, which can result in the same frequencies in large parent populations of random variables as the most probable combination of observations which the constraints can satisfy.