prepare working with uni server

src/models/desi/call_poly.py

+from POLY import POLY
+import sys
+import pathlib
+sys.path.append(str(pathlib.Path.absolute)+ '../../')
+from src.utils.slidingWindows import find_length_rank
+from src.run_model_wrapper import main
+
+#optimal hyperparameters from autors: 'POLY': {'periodicity': 1, 'power': 4}
+params = {
+        'periodicity': [1, 2, 3],
+        'power': [1, 2, 3, 4,5,7]
+}
+
+def run_POLY(data, periodicity=1, power=3, n_jobs=1):
+    slidingWindow = find_length_rank(data, rank=periodicity)
+    clf = POLY(power=power, window = slidingWindow)
+    clf.fit(data)
+    score = clf.decision_scores_
+    return score.ravel()
+
+model = 'POLY'
+output_path = '../../../docs/evaluation/'
+
+#writes results in .csv
+main(run_POLY,params,model,data_folders = '../../../data/', model_type='unsupervised',output_dir = output_path)
\ No newline at end of file

src/run_model_wrapper.py

@@ -47,16 +47,13 @@ Unsupervise_AD_Pool = ['FFT', 'SR', 'IForest', 'LOF', 'POLY', 'MatrixProfile', '
 Semisupervise_AD_Pool = ['SAND','OCSVM','AutoEncoder', 'CNN', 'LSTMAD', 'USAD', 'OmniAnomaly']
 '''
 
-def hyperparameter_optimization(run_model, data,label,train_data, hyperparams,slidingWindow, model_type = 'unsupervised'):
+def hyperparameter_optimization(run_model, data,label,train_data, hyperparams,slidingWindow,param_combinations, model_type = 'unsupervised'):
     #ignore warnings for hyperparameter tuning
     warnings.simplefilter("ignore", category=UndefinedMetricWarning)
     #warnings.simplefilter("ignore", category=np.RankWarning)
     best_params = None
     best_score = float("-inf")
 
-    # Creates combinations of all hyperparameters
-    param_combinations = list(itertools.product(*hyperparams.values()))
-
     #test each combination of hyperparameter and save best ones
     for param_set in param_combinations:
         params = dict(zip(hyperparams.keys(), param_set))
@@ -262,41 +259,40 @@ def main(run_model, hyperparams, model_name, data_folders = '../../../data/', mo
     path_data_all = data_folders + 'train/all'
     path_to_folder = data_folders + 'train/'
 
-    for folder in os.listdir(path_data_all):
-        #go through all files in train data and store additionally informations about group and anomaly types
-        grid_List = []
-        print(f"Start Processing files")
-        #reutrns [{file_name:str,train_data:list,data:list, label:list, sliding_window:int}]
-        path = os.path.join(path_data_all,folder)
-        file_data_dict_list = preprocess_data(path)     
+    #go through all files in train data and store additionally informations about group and anomaly types
+    grid_List = []
+    print(f"Start Processing files")
+    #reutrns [{file_name:str,train_data:list,data:list, label:list, sliding_window:int}]
+    file_data_dict_list = preprocess_data(path_data_all)    
 
-        print('Start Hyperparameter Tuning')
-        #find hyperparametrs for each file:
-        for time_series in file_data_dict_list:
-            filename = time_series['file_name']
-            data = time_series['data']
-            train_data = time_series['train_data']
-            label = time_series['label']
-            sliding_window = time_series['sliding_window']
+    print('Start Hyperparameter Tuning')
+    # Creates combinations of all hyperparameters
+    param_combinations = list(itertools.product(*hyperparams.values()))
+    #find hyperparametrs for each file:
+    for time_series in file_data_dict_list:
+        filename = time_series['file_name']
+        data = time_series['data']
+        train_data = time_series['train_data']
+        label = time_series['label']
+        sliding_window = time_series['sliding_window']
 
-            # Hyperparameter-Optimierung
-            best_params = hyperparameter_optimization(run_model, data,label,train_data, hyperparams,sliding_window,model_type)
-            print(f"File: {filename}, best hyperparameter: {best_params}")
-            #grid_list looks like this: [{ params: {key:value}, file_name:str,train_data:list,data:list, label:list, sliding_window:int}]
-            grid_List.append({'file_name':filename,'params':best_params,'train_data':train_data,'data':data,'label':label,'sliding_window':sliding_window})
+        # Hyperparameter-Optimierung
+        best_params = hyperparameter_optimization(run_model, data,label,train_data, hyperparams,sliding_window,param_combinations,model_type)
+        print(f"File: {filename}, best hyperparameter: {best_params}")
+        #grid_list looks like this: [{ params: {key:value}, file_name:str,train_data:list,data:list, label:list, sliding_window:int}]
+        grid_List.append({'file_name':filename,'params':best_params,'train_data':train_data,'data':data,'label':label,'sliding_window':sliding_window})
 
-        print('Start Training and evaluation')
-        results_list = train_test_model(run_model,grid_List,path_to_folder, model_type)
-        #convert list of dicts to a dict with lists
-        results = create_dict_from_array(results_list)
+    print('Start Training and evaluation')
+    results_list = train_test_model(run_model,grid_List,path_to_folder, model_type)
+    #convert list of dicts to a dict with lists
+    results = create_dict_from_array(results_list)
 
-        #check duration time
-        duration = sum(results['duration'])
-        print(f'Job done. Duration: {duration:.3f} seconds')
+    #check duration time
+    duration = sum(results['duration'])
+    print(f'Job done. Duration: {duration:.3f} seconds')
 
-        #store results for each model by folder 
-        results_df = pd.DataFrame(results)
-        output_csv = os.path.join(output_dir,model_name + folder + '.csv')
-        results_df.to_csv(output_csv, index=False)
-        print(f"Ergebnisse gespeichert in: {output_csv}")
-        time.sleep(60)
+    #store results for each model by folder 
+    results_df = pd.DataFrame(results)
+    output_csv = os.path.join(output_dir,model_name + '.csv')
+    results_df.to_csv(output_csv, index=False)
+    print(f"Ergebnisse gespeichert in: {output_csv}")