The counterargument raised by one the reviewers of this contribution was that it might be preferable to use one of the existing platforms for a machine learning course exercise rather than introduce yet another system and implement everything from scratch. The design goal of the BernoulliMix is to enable the student to concentrate on modeling and machine learning aspects in the term project, and therefore we should minimize the effort to learn a new environment or programming language. Matlab is widely known due to its adoption on the earlier mathematics courses as the tool. Other recognized programming tools among the students are quite varied, including various programming languages and mathematical programming environments. One environment that every student has been taught to use is the shell environment of the UNIX/Linux type of operating systems. This has guided us to introduce a command-line interface executed in a shell environment and the implementation to be such that it is easily compiled on a wide number of computing environments with minimal effort. Also, we tend to lean towards an open source implementation, which makes it possible to investigate and extend the program code, including benefits claimed by others [Sonnenburg et al., 2007]. The current implementation is written in C programming language [Kernighan & Ritchie, 1988]. This provides compatibility with a wide number of platforms and an efficient implementation, especially needed in the context of data mining. On top of the existing programs, it should be relatively easy to write interface functions for systems like Matlab [Mathworks, 1994] and R [R Project, 1997]. In order to be realistically extensible by all students, the documentation should also include a description of the application program interface (API).