A dynamic imputer that optimizes feature selection and segments subsets of data to maximize the number of input signal features.
This is from an unpublished paper I am able to make public from a non-disclosed project, the architecture of which is displayed below:
Parts of the projects have not been entirely generalized, this remains a pre-release version that is fully functional. Further updates will include better generalization of the imputer, improved modularization, as well as parallelization thanks to the functionality of the Numpy and Pandas libraries.
Feel free to contact me at [email protected].