This project shows how to implement the ABCDnn method used in an all-hadronic 4-top search to predict shape of backgrounds in data. In the code, the distribution of source is morphed to predict the distribution of the target in SR. The source and target should be provided in a flat rootfile format.

The Implement_ABCDnn.ipynb is main code. You can run it right away through Google Colab without any enviroment setup. There are some example root files in data directory so you can test the code.

Please see Instruction_for_ABCDnn.pdf for all detailed explanations.