make input models optional for hetero feature selection(#4661)

edit pipeline examples(#5008) Signed-off-by: Yu Wu <yolandawu131@gmail.com>
FederatedAI · Aug 11, 2023 · 5f6ee9d · 5f6ee9d
1 parent 8dd7c03
commit 5f6ee9d
Show file tree

Hide file tree

Showing 14 changed files with 78 additions and 80 deletions.
diff --git a/examples/pipeline/hetero_feature_binning/test_feature_binning_asymmetric.py b/examples/pipeline/hetero_feature_binning/test_feature_binning_asymmetric.py
@@ -36,7 +36,7 @@ def main(config="../config.yaml", namespace=""):
     psi_0 = PSI("psi_0")
     psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
 
     binning_0 = HeteroFeatureBinning("binning_0",
@@ -70,10 +70,10 @@ def main(config="../config.yaml", namespace=""):
     predict_pipeline = FateFlowPipeline()
 
     deployed_pipeline = pipeline.get_deployed_pipeline()
-    psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                  namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                     namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+                                                                                    namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+                                                                                       namespace=f"experiment{namespace}"))
 
     predict_pipeline.add_task(deployed_pipeline)
     predict_pipeline.compile()

diff --git a/examples/pipeline/hetero_feature_binning/test_feature_binning_bucket.py b/examples/pipeline/hetero_feature_binning/test_feature_binning_bucket.py
@@ -36,11 +36,11 @@ def main(config="../config.yaml", namespace=""):
     psi_0 = PSI("psi_0")
     psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
 
     psi_1 = PSI("psi_1")
-    psi_1.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+    psi_1.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
     psi_1.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
@@ -74,10 +74,10 @@ def main(config="../config.yaml", namespace=""):
     predict_pipeline = FateFlowPipeline()
 
     deployed_pipeline = pipeline.get_deployed_pipeline()
-    psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                  namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                     namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+                                                                                    namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+                                                                                       namespace=f"experiment{namespace}"))
 
     predict_pipeline.add_task(deployed_pipeline)
     predict_pipeline.compile()

diff --git a/examples/pipeline/hetero_feature_binning/test_feature_binning_quantile.py b/examples/pipeline/hetero_feature_binning/test_feature_binning_quantile.py
@@ -36,7 +36,7 @@ def main(config="../config.yaml", namespace=""):
     psi_0 = PSI("psi_0")
     psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
 
     binning_0 = HeteroFeatureBinning("binning_0",
@@ -69,10 +69,10 @@ def main(config="../config.yaml", namespace=""):
     predict_pipeline = FateFlowPipeline()
 
     deployed_pipeline = pipeline.get_deployed_pipeline()
-    psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                  namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                     namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+                                                                                    namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+                                                                                       namespace=f"experiment{namespace}"))
 
     predict_pipeline.add_task(deployed_pipeline)
     predict_pipeline.compile()

diff --git a/examples/pipeline/hetero_feature_selection/selection_testsuite.yaml b/examples/pipeline/hetero_feature_selection/selection_testsuite.yaml
@@ -38,7 +38,7 @@ tasks:
     script: test_feature_selection_binning.py
   selection-manual:
     script: test_feature_selection_manual.py
-  binning-statistics:
+  selection-statistics:
     script: test_feature_selection_statistics.py
-  binning-multi-model:
+  selection-multi-model:
     script: test_feature_selection_multi_model.py
diff --git a/examples/pipeline/hetero_feature_selection/test_feature_selection_binning.py b/examples/pipeline/hetero_feature_selection/test_feature_selection_binning.py
@@ -36,13 +36,12 @@ def main(config=".../config.yaml", namespace=""):
     psi_0 = PSI("psi_0")
     psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
 
     binning_0 = HeteroFeatureBinning("binning_0",
                                      method="quantile",
                                      n_bins=10,
-                                     bin_col=["x0"],
                                      transform_method="bin_idx",
                                      train_data=psi_0.outputs["output_data"]
                                      )
@@ -68,10 +67,10 @@ def main(config=".../config.yaml", namespace=""):
     predict_pipeline = FateFlowPipeline()
 
     deployed_pipeline = pipeline.get_deployed_pipeline()
-    psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                  namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                     namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+                                                                                    namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+                                                                                       namespace=f"experiment{namespace}"))
 
     predict_pipeline.add_task(deployed_pipeline)
     predict_pipeline.compile()

diff --git a/examples/pipeline/hetero_feature_selection/test_feature_selection_manual.py b/examples/pipeline/hetero_feature_selection/test_feature_selection_manual.py
@@ -36,11 +36,11 @@ def main(config=".../config.yaml", namespace=""):
     psi_0 = PSI("psi_0")
     psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
 
     selection_0 = HeteroFeatureSelection("selection_0",
-                                         method=["statistics"],
+                                         method=["manual"],
                                          train_data=psi_0.outputs["output_data"])
     selection_0.guest.component_setting(manual_param={"keep_col": ["x0", "x1"]})
     selection_0.hosts[0].component_setting(manual_param={"filter_out_col": ["x0", "x1"]})
@@ -60,10 +60,10 @@ def main(config=".../config.yaml", namespace=""):
     predict_pipeline = FateFlowPipeline()
 
     deployed_pipeline = pipeline.get_deployed_pipeline()
-    psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                  namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                     namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+                                                                                    namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+                                                                                       namespace=f"experiment{namespace}"))
 
     predict_pipeline.add_task(deployed_pipeline)
     predict_pipeline.compile()

diff --git a/examples/pipeline/hetero_feature_selection/test_feature_selection_multi_model.py b/examples/pipeline/hetero_feature_selection/test_feature_selection_multi_model.py
@@ -36,13 +36,12 @@ def main(config=".../config.yaml", namespace=""):
     psi_0 = PSI("psi_0")
     psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
 
     binning_0 = HeteroFeatureBinning("binning_0",
                                      method="quantile",
                                      n_bins=10,
-                                     bin_col=["x0"],
                                      transform_method="bin_idx",
                                      train_data=psi_0.outputs["output_data"]
                                      )
@@ -74,10 +73,10 @@ def main(config=".../config.yaml", namespace=""):
     predict_pipeline = FateFlowPipeline()
 
     deployed_pipeline = pipeline.get_deployed_pipeline()
-    psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                  namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                     namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+                                                                                    namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+                                                                                       namespace=f"experiment{namespace}"))
 
     predict_pipeline.add_task(deployed_pipeline)
     predict_pipeline.compile()

diff --git a/examples/pipeline/hetero_feature_selection/test_feature_selection_statistics.py b/examples/pipeline/hetero_feature_selection/test_feature_selection_statistics.py
@@ -36,7 +36,7 @@ def main(config=".../config.yaml", namespace=""):
     psi_0 = PSI("psi_0")
     psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
 
     statistics_0 = Statistics("statistics_0", input_data=psi_0.outputs["output_data"])
@@ -63,10 +63,10 @@ def main(config=".../config.yaml", namespace=""):
     predict_pipeline = FateFlowPipeline()
 
     deployed_pipeline = pipeline.get_deployed_pipeline()
-    psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                  namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                     namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+                                                                                    namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+                                                                                       namespace=f"experiment{namespace}"))
 
     predict_pipeline.add_task(deployed_pipeline)
     predict_pipeline.compile()

diff --git a/examples/pipeline/multi_model/test_multi.py b/examples/pipeline/multi_model/test_multi.py
@@ -16,7 +16,7 @@
 
 from fate_client.pipeline import FateFlowPipeline
 from fate_client.pipeline.components.fate import PSI, HeteroFeatureSelection, HeteroFeatureBinning, \
-    FeatureScale, Union, DataSplit, CoordinatedLR, CoordinatedLinR, Statistics, Sample, Evaluation
+    FeatureScale, Union, DataSplit, CoordinatedLR, Statistics, Sample, Evaluation
 from fate_client.pipeline.interface import DataWarehouseChannel
 from fate_client.pipeline.utils import test_utils
 
@@ -38,7 +38,7 @@ def main(config="../config.yaml", namespace=""):
     psi_0 = PSI("psi_0")
     psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
 
     data_split_0 = DataSplit("data_split_0", input_data=psi_0.outputs["output_data"],
@@ -109,7 +109,7 @@ def main(config="../config.yaml", namespace=""):
     predict_pipeline = FateFlowPipeline()
 
     deployed_pipeline = pipeline.get_deployed_pipeline()
-    psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+    psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
     psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))

diff --git a/examples/pipeline/scale/test_scale_min_max.py b/examples/pipeline/scale/test_scale_min_max.py
@@ -36,11 +36,11 @@ def main(config="../config.yaml", namespace=""):
     psi_0 = PSI("psi_0")
     psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
 
     psi_1 = PSI("psi_1")
-    psi_1.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+    psi_1.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
                                                                   namespace=f"experiment{namespace}"))
     psi_1.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
                                                                      namespace=f"experiment{namespace}"))
@@ -57,7 +57,7 @@ def main(config="../config.yaml", namespace=""):
 
     statistics_0 = Statistics("statistics_0",
                               metrics=["max", "min", "mean", "std"],
-                              input_data=feature_scale_1.outputs["train_output_data"])
+                              input_data=feature_scale_1.outputs["test_output_data"])
 
     pipeline.add_task(psi_0)
     pipeline.add_task(psi_1)
@@ -77,10 +77,10 @@ def main(config="../config.yaml", namespace=""):
     predict_pipeline = FateFlowPipeline()
 
     deployed_pipeline = pipeline.get_deployed_pipeline()
-    psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                  namespace=f"experiment{namespace}"))
-    psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
-                                                                     namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.guest.component_setting(input_data=DataWarehouseChannel(name="breast_hetero_guest",
+                                                                                    namespace=f"experiment{namespace}"))
+    deployed_pipeline.psi_0.hosts[0].component_setting(input_data=DataWarehouseChannel(name="breast_hetero_host",
+                                                                                       namespace=f"experiment{namespace}"))
 
     predict_pipeline.add_task(deployed_pipeline)
     predict_pipeline.compile()