Sequential Good-Turing and the Missing Species Problem

Abstract: This essay introduces the sequential Good-Turing estimator and reviews the Good-Turing, Good-Toulmin and smoothed Good-Toulmin estimators. Some theoretical properties and drawbacks of the estimators are described. MonteCarlo simulation is then used to compare the performance of the sequential Good-Turing estimator to the performance of the Good-Toulmin estimator along with the smoothed Good-Toulmin estimator, on both real and simulated data. In certain scenarios the Monte-Carlo method outperforms the smoothed Good-Toulmin estimator.