This is nice, thanks for posting this. we are building an opensource datascience tool with similar vision as yours, and could certainly use something like this as initial data wrangling and preparation stage. https://github.com/rebataur/rapidium