Featuresstats++ is advanced, comprehensive statistical software, with included C++ source code.
stats+++ is advanced, comprehensive statistical software for: data collection and preprocessing, statistics, machine learning, and optimization.
In addition to direct use in its executable version, its C++ source code is included. This means that additional features/modifications are easy to implement, making it suitable for academic and government research. Some of the features of stats++ are highlighted below.
- Data collection and preprocessing
- Machine learning
- Data postprocessing
Full list of features.
stats++ also provides many advanced features, that one may not expect from open-source software.
As an example, an image of a restricted Boltzmann machine (a building block of deep-learning architectures) is shown to the left.
Simple User Interface
stats++ compiles into a set of executable version.
In this version, all parameters can be set using a simple input file, or from the command line (see the images to the right, for example). Though, most of these are automatically optimized and specified by stats++, for simplicity and robustness.
stats++ has been designed from the ground up to be highly scalable, in order to take advantage of the speedup offered by high-performance computing.
Based on the Message Passing Interface (MPI) system, stats++ is capable of performing simulations on systems ranging from personal computers to massively parallel systems. An example of the Titan supercomputer is shown to the right. There is no limit to the size of simulation that can be performed, in terms of time, memory, etc.
In development is the capability for stats++ to take advantage of advanced architectures (GPUs, co-processors, etc.).