add initial pipeline to multiobjective test

aimclub · Jul 12, 2023 · a0fcb5f · a0fcb5f
1 parent 058a0f7
commit a0fcb5f
Showing 1 changed file with 21 additions and 10 deletions.
diff --git a/test/integration/quality/test_quality_improvement.py b/test/integration/quality/test_quality_improvement.py
@@ -5,6 +5,8 @@
 import numpy as np
 
 from fedot.api.main import Fedot
+from fedot.core.pipelines.node import PipelineNode
+from fedot.core.pipelines.pipeline import Pipeline
 from fedot.core.utils import fedot_project_root
 
 warnings.filterwarnings("ignore")
@@ -19,13 +21,13 @@ def test_classification_quality_improvement():
     problem = 'classification'
     with_tuning = False
 
-    common_params= dict(problem=problem,
-                        n_jobs=1,
-                        use_pipelines_cache=False,
-                        use_preprocessing_cache=False,
-                        with_tuning=with_tuning,
-                        logging_level=logging.DEBUG,
-                        seed=seed)
+    common_params = dict(problem=problem,
+                         n_jobs=1,
+                         use_pipelines_cache=False,
+                         use_preprocessing_cache=False,
+                         with_tuning=with_tuning,
+                         logging_level=logging.DEBUG,
+                         seed=seed)
 
     expected_baseline_quality = 0.750
     baseline_model = Fedot(**common_params)
@@ -56,13 +58,22 @@ def test_multiobjective_improvement():
     metrics = [quality_metric, complexity_metric]
 
     timeout = 2
-    composer_params = dict(num_of_generations=10,
+    composer_params = dict(num_of_generations=5,
                            pop_size=3,
                            with_tuning=False,
-                           preset='fast_train',
+                           preset='best_quality',
                            metric=metrics)
 
+    initial_pipeline = Pipeline(
+        PipelineNode('logit',
+                     nodesfrom=[
+                         PipelineNode('rf', nodes_from=[PipelineNode('rf')]),
+                         PipelineNode('rf', nodes_from=[PipelineNode('rf')])
+                     ])
+    )
+
     auto_model = Fedot(problem=problem, timeout=timeout, seed=seed, logging_level=logging.DEBUG,
+                       initial_assumption=initial_pipeline,
                        **composer_params, n_jobs=1, use_pipelines_cache=False, use_preprocessing_cache=False)
     auto_model.fit(features=train_data_path, target='target')
     auto_model.predict_proba(features=test_data_path)
@@ -71,7 +82,7 @@ def test_multiobjective_improvement():
     quality_improved, complexity_improved = check_improvement(auto_model.history)
 
     assert auto_metrics[quality_metric] > 0.75
-    assert auto_metrics[complexity_metric] >= 0.2
+    assert auto_metrics[complexity_metric] <= 0.2
     assert quality_improved
     assert complexity_improved