Math.NET Numerics is a pretty good open source library for .NET. According to its site:
Numerics is the numerical foundation of the Math.NET project, aiming to provide methods and algorithms for numerical computations in science, engineering and every day use. Covered topics include special functions, linear algebra, probability models, random numbers, interpolation, integral transforms and more.
If commercial library is an acceptable choice for you there are several of them I can recommend.
- NMath Stats from CenterSpace Software is a good choice. This library is specially designed for statistical computing and it is promised to be very efficient.
- FinMath is rather new but promising library. It contains a lot of statistical distributions, routines for computing descriptive statistics, a few variants of least squares data fitting algorithms (OLS, GLS etc.), PCA, ICA and other implementations of various algorithms and techniques.