Example: Distributed model deployment and loading of td_lightgbm sklearn model trained in Vantage - Teradata Package for Python

>>> obj_skl_m = td_lightgbm.LGBMModel(num_leaves=15, objective="binary", n_estimators=5)

>>> obj_skl_m.fit(df_x_classif, df_y_classif, sample_weight=df_train_classif.select("group_column"),
                  partition_columns=["partition_column_1", "partition_column_2"],
                  callbacks=[td_lightgbm.log_evaluation()])

   partition_column_1  partition_column_2                                                         model
0                   1                  11  LGBMModel(n_estimators=5, num_leaves=15, objective='binary')
1                   0                  11  LGBMModel(n_estimators=5, num_leaves=15, objective='binary')
2                   1                  10  LGBMModel(n_estimators=5, num_leaves=15, objective='binary')
3                   0                  10  LGBMModel(n_estimators=5, num_leaves=15, objective='binary')

>>> obj_deploy_skl_m = obj_skl_m.deploy(model_name="lightgbm_sklearn_multi_model")

Model is saved.

>>> obj_skl_load_model_m = td_lightgbm.load("lightgbm_sklearn_multi_model")

>>> obj_skl_load_model_m.predict(df_x_classif, raw_score=True, pred_leaf=True)

partition_column_1	partition_column_2				col1  ...  			col4  lgbmmodel_predict_1  lgbmmodel_predict_2  ...   lgbmmodel_predict_5
				 1					10	0.05870288650510  ...  -1.17785382356  					0  					 0  ...  					1
				 1					10	-1.1615345341513  ...  -0.86423207141  					3  					 2  ...  					1
				 1					10	0.78433767392278  ...  0.720943051314  					1  					 1  ...  					0
				 1					10	-0.2464757122084  ...  -0.40166606892  					8  					 3  ...  					2
				 1					10	0.23164855319438  ...  -1.26023579265  					0  					 0  ...  					1
				 1					10	-0.7267611459507  ...  -0.32066543120  					7  					 1  ...  					2
				 0					10	2.08263134345229  ...  1.258519751162  					2  					 2  ...  					2
				 0					10	1.43975694528171  ...  0.656882820481  					3  					 3  ...  					3
				 0					10	-0.6124690179513  ...  -0.48785374115  					5  					 3  ...  					3
				 0					10	1.30373436256498  ...  0.933048019921  					2  					 2  ...  					2

>>> obj_skl_load_model_m.feature_importances_

	partition_column_1	partition_column_2	 feature_importances_
0					 1					11			[3, 11, 1, 0]
1					 0					11			 [0, 5, 7, 0]
2					 1					10			 [2, 7, 2, 2]
3					 0					10			[4, 10, 0, 1]