automl-docker

based on rootproject/root:6.22.02-ubuntu20.04

docker and script to test some automl packages for binary classification

Installation:

cd automl-docker

ln -s requirements_ag.txt requirements.txt (or, alternaively, requirements_askl.txt)

docker build -t my-ml -f Dockerfile .

Usage

docker run --rm -it --shm-size=3.00gb -v $PWD/workdir:/home/foo/workdir --user $(id -u) my-ml

Train the model

foo:~$ python3 -m pip freeze

foo:~$ cd /home/foo/workdir/

foo:~$ python3 -i etahml.py

Use the model in a C++ program

foo:~$ mkdir build

foo:~$ cd build

foo:~$ cmake ..

foo:~$ make

After placing the input .root file in build folder, copy model and run the program:

foo:~$ cp ../etaPi_BDT_clf.joblib .

foo:~$ ./example