nnetsauce is a Python package for Statistical/Machine learning built on top of Numpy, Scipy, and scikit-learn under the BSD license. In nnetsauce, pattern recognition is achieved by combining single-layer networks (SLNN). These SLNN building blocks constitute the basis of many custom models that can be built by the user, including models with deeper learning architectures.
Examples of use, for illustration: nnetsauce can (for example) be trained to distinguish between malignant or benign tumors, depending on their characteristics (see examples on Github), with a certain degree of confidence . Or, by using your historical data of income and expenditures, it can be trained to forecast your savings in the next months.
The package is very new. For those who want to try it, give feedback, make it available in R, contribute to the development on Github, feel free to jump in.