Fitting Mixtures of Gaussian Distributions

This webinar describes new procedures added to Statgraphics 19 for fitting mixtures of Gaussian distributions to both univariate and bivariate data. Mixture distributions are useful for fitting data that have more than one mode and for data that are generated by more than one mechanism. The procedures use a numerical algorithm to estimate the values of the means, standard deviations and covariance (for bivariate data). As an example, distributions are fit to data describing the length of eruptions of the Old Faithful geyser at Yellowstone National Park and the waiting time until the next eruption.