PantherPro20/Decision-Tree-from-Scratch---No-Scikit-Learn ? reverse-engineered prompt

Build me a simple decision tree classifier in Python from scratch, without using scikit learn.

I want it to train on a small dataset in the repo, split the data, learn the tree, and make predictions so I can see the accuracy at the end. Please keep it easy to follow, since I want to understand how the tree is built step by step, not just use a library.

If there are any bugs or rough edges in the current notebook or code files, clean them up and make the project run nicely from start to finish. It would be great to have the main logic in a clear Python file, with the notebook showing how to use it and test it.

Try to keep the result simple and readable, and if you need to check any current Python docs or best practices online, go ahead and do that.