Starting to add a first version of weight density plot for stage4

code/compute_results.py

@@ -28,7 +28,6 @@ def extract_scores_across_seeds_and_extracted_forest_sizes(models_dir, results_d
     experiment_train_scores = dict()
     experiment_dev_scores = dict()
     experiment_test_scores = dict()
-    experiment_weights = dict()
     all_extracted_forest_sizes = list()
 
     # Used to check if all losses were computed using the same metric (it should be the case)
@@ -45,7 +44,6 @@ def extract_scores_across_seeds_and_extracted_forest_sizes(models_dir, results_d
         experiment_train_scores[seed] = list()
         experiment_dev_scores[seed] = list()
         experiment_test_scores[seed] = list()
-        experiment_weights[seed] = list()
 
         # List the forest sizes in models/{experiment_id}/seeds/{seed}/extracted_forest_sizes
         extracted_forest_sizes = os.listdir(extracted_forest_sizes_root_path)
@@ -66,8 +64,6 @@ def extract_scores_across_seeds_and_extracted_forest_sizes(models_dir, results_d
             experiment_test_scores[seed].append(model_raw_results.test_score)
             # Save the metric
             experiment_score_metrics.append(model_raw_results.score_metric)
-            # Save the weights
-            #experiment_weights[seed].append(model_raw_results.model_weights)
 
     # Sanity checks
     if len(set(experiment_score_metrics)) > 1:
@@ -76,7 +72,7 @@ def extract_scores_across_seeds_and_extracted_forest_sizes(models_dir, results_d
         raise ValueError("The extracted forest sizes aren't the sames across seeds.")
 
     return experiment_train_scores, experiment_dev_scores, experiment_test_scores, \
-        all_extracted_forest_sizes[0], experiment_score_metrics[0]#, experiment_weights
+        all_extracted_forest_sizes[0], experiment_score_metrics[0]
 
 def extract_scores_across_seeds_and_forest_size(models_dir, results_dir, experiment_id, extracted_forest_sizes_number):
     experiment_id_path = models_dir + os.sep + str(experiment_id) # models/{experiment_id}
@@ -123,6 +119,37 @@ def extract_scores_across_seeds_and_forest_size(models_dir, results_dir, experim
 
     return experiment_train_scores, experiment_dev_scores, experiment_test_scores, experiment_score_metrics[0]
 
+def extract_weights_across_seeds(models_dir, results_dir, experiment_id):
+    experiment_id_path = models_dir + os.sep + str(experiment_id) # models/{experiment_id}
+    experiment_seed_root_path = experiment_id_path + os.sep + 'seeds' # models/{experiment_id}/seeds
+    experiment_weights = dict()
+
+    # For each seed results stored in models/{experiment_id}/seeds
+    seeds = os.listdir(experiment_seed_root_path)
+    seeds.sort(key=int)
+    for seed in seeds:
+        experiment_seed_path = experiment_seed_root_path + os.sep + seed # models/{experiment_id}/seeds/{seed}
+        extracted_forest_sizes_root_path = experiment_seed_path + os.sep + 'extracted_forest_sizes' # models/{experiment_id}/seeds/{seed}/forest_size
+
+        # {{seed}:[]}
+        experiment_weights[seed] = list()
+
+        # List the forest sizes in models/{experiment_id}/seeds/{seed}/extracted_forest_sizes
+        extracted_forest_sizes = os.listdir(extracted_forest_sizes_root_path)
+        extracted_forest_sizes = [nb_tree for nb_tree in extracted_forest_sizes if not 'no_weights' in nb_tree ]
+        extracted_forest_sizes.sort(key=int)
+        all_extracted_forest_sizes.append(list(map(int, extracted_forest_sizes)))
+        for extracted_forest_size in extracted_forest_sizes:
+            # models/{experiment_id}/seeds/{seed}/extracted_forest_sizes/{extracted_forest_size}
+            extracted_forest_size_path = extracted_forest_sizes_root_path + os.sep + extracted_forest_size
+            # Load models/{experiment_id}/seeds/{seed}/extracted_forest_sizes/{extracted_forest_size}/model_raw_results.pickle file
+            model_raw_results = ModelRawResults.load(extracted_forest_size_path)
+            # Save the weights
+            experiment_weights[seed].append(model_raw_results.model_weights)
+
+    return experiment_weights
+
+
 if __name__ == "__main__":
     # get environment variables in .env
     load_dotenv(find_dotenv('.env'))
@@ -400,6 +427,10 @@ if __name__ == "__main__":
             xlabel='Number of trees extracted',
             ylabel=experiments_score_metric,
             title='Loss values of {}\nusing best params of previous stages'.format(args.dataset_name))
+
+        experiment_weights = extract_weights_across_seeds(args.models_dir, args.results_dir, args.experiment_ids[2])
+
+        Plotter.weight_density(experiment_weights, os.path.join(output_path, 'weight_density.png'))
     elif args.stage == 5:
         # Retreive the extracted forest sizes number used in order to have a base forest axis as long as necessary
         extracted_forest_sizes_number = retreive_extracted_forest_sizes_number(args.models_dir, args.experiment_ids[1])