Why Choose Mathematica for Your Statistics Analysis?

Like any statistics package, Mathematica provides a numerical and graphical toolset to illustrate, simulate, and find approximate numeric solutions to numerical problems. What is special in Mathematica is the powerful symbolic/algebraic engine that allows you to solve complicated symbolic problems. In addition, in the Mathematica programming environment you are not restricted to a limited number of preexisting functions but can develop your own supplementary commands, creating and refining the exact test or model that is relevant for the particular problem you are interested in.

Numerical Accuracy

Mathematica provides an arbitrary-precision numeric engine, whereas most software packages provide only finite-precision numerics. Hence, if accuracy is important to your work, Mathematica is your natural choice.

The National Institute of Standards and Technology (NIST) created the Statistical Reference Datasets (StRD), a set of numerical benchmark values for linear regression, nonlinear regression, univariate summary statistics, and one-way analysis of variance. According to B. D. McCullough,¹ no software package was able to match the accuracy of these tests. However, one year later, when subjecting Mathematica to an identical battery of tests, the same author² found that Mathematica's accuracy easily outperformed all of the competition in each of the four benchmark areas.

A Complete Environment for Both Computations and Technical Writing

Mathematica documents are platform-independent ASCII files called notebooks. Mathematica notebooks enable you to incorporate text, pictures, equations, animations, and computations into a single interactive, live document. This document is easy to distribute to anyone to whom you want to report your results or from whom you wish to get comments. A notebook can also be exported as a TeX, XML, or HTML document or as one of the many other forms supported by Mathematica.

Moreover, Mathematica runs on a wide variety of platforms, making it easy to share your work with people working in different institutions and possibly using different platforms. This is especially valuable in academia, where coauthorship is common.

¹McCullough, B. D. "Assessing the Reliability of Statistical Software: Part II," The American Statistician 53, no.2 (1999) 149-159.

²McCullough, B. D. "The Accuracy of Mathematica 4.0 as a Statistical Package," Computational Statistics 15, no. 2 (2000) 279-299.