Automunge is an open source platform for preparing tabular data for machine learning. For complete documentation please see our READ ME on GitHub: https://t.co/AICGEbZYKc Or for our website including links to essays and contact information please visit: https://t.co/AzVUxXJy5R

As clarification, there are a small number of optional imports for supplemental operations that I do not believe are surveyed by these Snyk audits, which I think only looks at default imports. Comprehensive import options are documented in the readme.

Celebrating approximately 2+ years of zero known vulnerabilities to the Automunge library and associated dependencies. Thanks @snyksec

as well as a crisply defined method for defining sets of custom univariate data transformations as fit to a basis, in an open source setting. Although the pace of rollouts were fairly frequent, they were done so with attempted comprehensive validations and backward compatibility.

At the time it was developed, although there were countless data wrangling products on the market both paid and open source, the differentiation of the automunge project, amongst other things, included integrated ML infill allowing to mitigate channels for data leakage (1/2)

@_NicT_ @IEEE @UF The primary source of the founder @_nict_’s income through this period was from funding his startup with his life savings from preceding career in engineering, and now supplemented by investing income by his long term holdings. He does not follow or participate in cryptocurrency.

@_NicT_ @IEEE @UF The founder @_nict_ is operating under the assumption that although he is a US citizen, there were family immigration matters that may have been an enormous complicating factor and is grateful that they appear to be at least close to resolved.

@_NicT_ @IEEE @UF So far, the primary meaningful opportunities for feedback have been associated with submittals to peer reviewed academic conference venues, for which the founder @_NicT_ had several workshop papers distributed in a few venues like @NeurIPS. He has recently initiated a new project

@_NicT_ @IEEE @UF The founder @_nict_ now resides in Orlando, and is operating on assumption that the work of the @automunge library is in “limbo” through an unknown channel of oversight which has somehow obstructed his ability to conduct any form of business without due processes during this 8yrs

@_NicT_ @IEEE After securing patent @_NicT_ moved to Gainesville, FL and spent a year attempting to secure an audience with an appropriate researcher to solicit feedback on his work and research which he considered appropriate as a three time @UF alumni and lifetime alumni association member.

@_NicT_ @IEEE After withdrawing from that working group without traction the founder spent an additional year self prosecuting the remaining IP with the hope that the patent that was eventually awarded in January 2024 could open the door to consulting revenue for the founder @_NicT_.

After pivoting from the project the founder @_NicT_ spent the following year volunteering for an @IEEE working group with the hope that if he could provide enough value to the participating Fortune 500 tech firms it could open the door to consulting income FYI.

The business model of Automunge library was partly built on assumption that a sufficiently useful or at least insightful open source implementation could open the door to consulting income to the founder @_nict_ FYI.

Either my work with the automunge library was trivial and there should be no hesitancy to cite related to scope of the awarded patent, or my open source work was of great importance and should be recognized privately if not publicly. Can’t have it both ways by sweeping under rug.

For avoidance of doubt:

(Reverted a code edit in GitHub repo that appeared to introduce a large number of deleted lines in the repo, was unknown until now.)

As further clarification, the patent isn’t on the API, the patent is associated with the documented process that is built around the abstractions used as basis for the API.

FYI

The automunge patent wasn’t an attempt to own tabular preprocessing for supervised learning and whatnot, it was awarded based on demonstrating novelty of the abstractions used as basis for the API. (The claim 1 from which other claims are dependent.) The “family tree primitives”.

In case there are ever any connectivity issues with the (automunge dot com) homepage, the internet archive service (archive dot org) has been indexing the website for a while.

I should clarify that the “stochastic perturbations” paper made note of an option for sampling entropy from external resources making use of the QRAND library. I cannot vouch for this resource, it was merely used based on convenient interface for accessing np.random with qiskit.