Meta.Numerics.Statistics Namespace

This namespace contains types for doing basic and advanced statistics. It contains APIs for finding moments and percentiles, measuring associations, fitting to models, and other statistical operations.

The central class for operating on samples consisting of independent measurements of a single variable is the Univariate class. The central class for operating on samples of independent bivariate data (i.e. paired measurements) is the Bivariate class. The central class for operating on samples of independent measurements of multiple variables is the Multivariate class. The central class for operating on time series data is the Series class.

All of these central classes consist of static methods that accept one or more columns of data. Each column can be of any type that implements the appropriate collection interface (e.g. IReadOnlyListT). Many of the methods are extension methods, so they effectively become instance methods on all such types.

The data storage classes Sample, BivariateSample, MultivariateSample, and TimeSeries, which appeared in earlier versions of Meta.Numerics, have been removed. All the summary statistics, statistical tests, fits, and other functionality that they provided can now be accessed using the methods of the Univariate, Bivariate, Multivariate and Series classes. The new methods can be applied to data in arbitrary collection types, so specialized storage types are no longer necessary. We expect to remove the storage types in a future release.