Chapter 4. Experiment Management in MLflow


$ cd ~/tmp
$ source stockpred_env/bin/activate


$ cd ~/tmp/Machine-Learning-Engineering-with-MLflow/Chapter04/gradflow/
$ make


$ docker ps
CONTAINER ID   IMAGE                               COMMAND                  CREATED          STATUS          PORTS                                       NAMES
3d65aa200d03   gradflow/workbench/jupyter:0.1.0    "tini -g -- start-no…"   19 seconds ago   Up 17 seconds   0.0.0.0:8888->8888/tcp, :::8888->8888/tcp   gradflow-jupyter-1
0ba4627c9518   gradflow/workbench/mlflow:0.1.0     "/bin/sh -c './wait-…"   19 seconds ago   Up 18 seconds   0.0.0.0:5000->5000/tcp, :::5000->5000/tcp   gradflow-mlflow-1
fed2e90e166f   gradflow/workbench/postgres:0.1.0   "docker-entrypoint.s…"   20 seconds ago   Up 18 seconds   0.0.0.0:5432->5432/tcp, :::5432->5432/tcp   gradflow-postgres-1


http://localhost:5000


http://localhost:8888

Chapter04/gradflow/notebooks/retrieve_training_data.ipynb


Будем запускать эксперименты

• Logistic Classifier (Chapter04/gradflow/notebooks/mlflow_run_logistic_regression.ipynb) • Xgboost (Chapter04/gradflow/notebooks/mlflow_run_xgboost.ipynb) • Keras (Chapter04/gradflow/notebooks/mlflow_run_keras.ipynb)


Chapter04/gradflow/notebooks/hyperopt_optimization_logistic_regression_mlflow.ipynb