Enable pruning when node error rate equals children error rate
I agree with the assessment in issue #1, here's a very small PR with the appropriate patch
Coverage remained the same at 93.892% when pulling b210973a7e6ff184d291a38944a86adc7ee39b09 on pipsqueaker:pipsqueaker-patch-1 into f6ab4a749a7303d49e31e3ffa4253453e93e99b9 on svaante:master.
I surely agree with #1 but somehow i haven't found the time to maintain this repo. I would like to commend you for your very clean commit history.
I will contact @ONordander as he is the owner of the PIP repository, as your commit contain a version update. If he as time to update the version then I will merge this pull request.
I would like to give some background history to why "=>" where used over ">". This only started because the java lib WEKA implemented the J48 algorithm which is a open source implementation of c4.5 algorithm which is an extension of ID3 algorithm and WEKA J48 was used extensively early on during testing and it was concluded that J48 used "=>" and therefore it was implemented that way. Sorry for ranting but I felt the need to document some history behind this project.