README.md

mlflow-examples - xgboost

Overview

• XBGoost with sklearn train and predict.
• Saves model in xgboost format.
• Wine quality dataset train/wine-quality-white.csv.

Training

python train.py --experiment_name xgboost --estimators 20000 --max_depth 5 

mlflow run . --experiment_name xgboost -P estimators=20000 -P max_depth=5 

Predictions

Score with mlflow.xgboost.load_model and mlflow.pyfunc.load_model. You can either use a runs or models URI.

python predict.py runs:/7e674524514846799310c41f10d6b99d/xgboost-model

python predict.py models:/xgboost_wine/production

=== mlflow.xgboost.load_model
model: <xgboost.core.Booster object at 0x113678b70>
predictions: [5.3752966 5.2566967 5.4596467 ... 5.347645  6.682991  6.0259304]

=== mlflow.pyfunc.load_model
model: <mlflow.xgboost._XGBModelWrapper object at 0x10e9eb198>
predictions: [5.3752966 5.2566967 5.4596467 ... 5.347645  6.682991  6.0259304]