uniovi-simur-wearablepermed-ml 1.1.0__py3-none-any.whl → 1.2.0__py3-none-any.whl

{uniovi_simur_wearablepermed_ml-1.1.0.dist-info → uniovi_simur_wearablepermed_ml-1.2.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: uniovi-simur-wearablepermed-ml
-Version: 1.1.0
+Version: 1.2.0
 Summary: Uniovi Simur WearablePerMed Machine Learning.
 Home-page: https://github.com/Simur-project/uniovi-simur-wearablepermed-ml.git
 Author: Miguel Salinas Gancedo

uniovi_simur_wearablepermed_ml-1.2.0.dist-info/RECORD

@@ -0,0 +1,19 @@
+uniovi_simur_wearablepermed_ml-1.2.0.dist-info/licenses/LICENSE.txt,sha256=MJSf2LY2uh50n0Y7vPzbMcIfTXiY_IvEp7dQMhSHBWo,1089
+wearablepermed_ml/__init__.py,sha256=Jl6IIVd6LQLVAQG-uKC7nnY2204vk_YrBMz3bC4JxvU,601
+wearablepermed_ml/run_trainer_and_tester_30_times.py,sha256=rYa9vyOF-cqjs61_FqqszimBVwU8WDg-ydTHpAponFQ,8788
+wearablepermed_ml/tester.py,sha256=rh_ku_upI9jtHjr9GtEpj-q5G6fYN-CFv7MloYqcLUY,4364
+wearablepermed_ml/trainer.py,sha256=CUzRPCK6NEujS6mFLKspY6HA7I-V0DXVtRucFrO4V1Q,43745
+wearablepermed_ml/basic_functions/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+wearablepermed_ml/basic_functions/address.py,sha256=VYAvZLqyd4DvX2bYUJI5fnepKPMaI4CY7uiaeCsNauM,664
+wearablepermed_ml/data/DataReader.py,sha256=Bd1AqSytx0AXx1S_C9osk9I205vXdoNYP6IZctDSOmw,18476
+wearablepermed_ml/data/__init__.py,sha256=ce3QTcR18knSsPJguwJkVqHJnTnxYv4WWBKngmTL_BI,58
+wearablepermed_ml/models/SiMuR_Model.py,sha256=WXAkvFhOhLkzgCjiwkmT4X9693-zlolnhsBepNCXarc,30346
+wearablepermed_ml/models/__init__.py,sha256=HkTS7w1Mpj0TP7ACtgcKv2gZEVYDNqAectiwGBI_Mmo,108
+wearablepermed_ml/models/model_generator.py,sha256=8RX5HZWcz1aZV5hl5RNRchihrlNYmXV8LgAD2UrPsxw,1906
+wearablepermed_ml/testing/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+wearablepermed_ml/testing/testing.py,sha256=93ki6WocLcY5MpKHOscYOqm0J1A3FGXjEKCSDwOmr00,10068
+uniovi_simur_wearablepermed_ml-1.2.0.dist-info/METADATA,sha256=HJS303mn7jlsDyuaI_pK7N9HVOkpS2r3Q8Jdw8yCgoo,15626
+uniovi_simur_wearablepermed_ml-1.2.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+uniovi_simur_wearablepermed_ml-1.2.0.dist-info/entry_points.txt,sha256=BW3Dw-fW_hquNHbK1q7pUNgIGwRDMPHqs9znQE6SECM,96
+uniovi_simur_wearablepermed_ml-1.2.0.dist-info/top_level.txt,sha256=PK7Cm_vvudFpRS-vGYiQDsZLkFZ6mdaocJqSPXHMP4c,18
+uniovi_simur_wearablepermed_ml-1.2.0.dist-info/RECORD,,

wearablepermed_ml/data/__init__.py

@@ -1 +1,2 @@
+# export python modules
 from .DataReader import DataReader

wearablepermed_ml/models/model_generator.py

@@ -1,7 +1,7 @@
-from models import *
-from basic_functions.address import *
+from wearablepermed_ml.models import *
+from wearablepermed_ml.basic_functions.address import *
 import pandas as pd
-from data import DataReader
+from wearablepermed_ml.data import DataReader
 import ast
 
 # Model factory pattern

wearablepermed_ml/run_trainer_and_tester_30_times.py

@@ -8,10 +8,10 @@ N_RUNS = 30                                                  # Número de ejecuc
 # Ruta Windows
 # case_id_folder = "D:\\DATA_PMP_File_Server\\output"          # Carpeta base de los datos
 # Ruta Linux
-# case_id_folder = "/mnt/nvme1n2/git/uniovi-simur-wearablepermed-data/output"
-case_id_folder = "/mnt/simur-fileserver/data/wearablepermed/output"
+case_id_folder = "/mnt/nvme1n2/git/uniovi-simur-wearablepermed-data/output"
+# case_id_folder = "/mnt/simur-fileserver/data/wearablepermed/output"
 
-case_id = "cases_dataset_PI_M/case_PI_M_BRF_superclasses"                                          # Identificador del caso
+case_id = "cases_dataset_PI_M/case_PI_M_BRF_superclasses_activity_intensity/"                                          # Identificador del caso
 
 # Argumentos para el script de entrenamiento
 train_args = [
@@ -24,7 +24,8 @@ train_args = [
     "--ml-models", "RandomForest",                                  # Modelo ML a usar
     "--training-percent", "70",                              # Porcentaje de datos para entrenamiento
     # "--validation-percent", "20",                            # Porcentaje de datos para validación
-    "--create-superclasses"                                  # Flag opcional para crear superclases
+    # "--create-superclasses"                                  # Flag opcional para crear superclases
+    "--create-superclasses-CPA-METs"
 ]
 
 # Argumentos para el script de test
@@ -38,7 +39,8 @@ test_args = [
     "--model-id", "RandomForest",                                   # Modelo ML usado para test
     "--training-percent", "70",                              # Porcentaje usado en entrenamiento
     # "--validation-percent", "20",                            # Porcentaje de datos para validaciones
-    "--create-superclasses"                                  # Flag opcional
+    # "--create-superclasses"                                  # Flag opcional
+    "--create-superclasses-CPA-METs"
 ]
 
 # Ruta del ejecutable de Python del entorno virtual (Windows)

wearablepermed_ml/tester.py

@@ -4,7 +4,7 @@ import sys
 import argparse 
 import logging
 
-from testing import testing
+from wearablepermed_ml.testing import testing
 
 __author__ = "Miguel Salinas <uo34525@uniovi.es>, Alejandro <uo265351@uniovi.es>"
 __copyright__ = "Uniovi"

wearablepermed_ml/testing/testing.py

@@ -1,8 +1,8 @@
 from enum import Enum
 import json
-from data import DataReader
-from models.model_generator import modelGenerator
-from basic_functions.address import *
+from wearablepermed_ml.data import DataReader
+from wearablepermed_ml.models.model_generator import modelGenerator
+from wearablepermed_ml.basic_functions.address import *
 import numpy as np
 import matplotlib.pyplot as plt
 import seaborn as sns
@@ -104,18 +104,22 @@ def tester(case_id_folder, model_id, create_superclasses, create_superclasses_CP
 
     # testing the model
     y_predicted_train = model.predict(model.X_train)
-    y_predicted_validation = model.predict(model.X_validation)
     y_predicted_test = model.predict(model.X_test)
 
     # get the class with the highest probability
     if (model_id == ML_Model.ESANN.value or model_id == ML_Model.CAPTURE24.value):
+        y_predicted_validation = model.predict(model.X_validation)
         y_final_prediction_train = np.argmax(y_predicted_train, axis=1)
         y_final_prediction_validation = np.argmax(y_predicted_validation, axis=1)
         y_final_prediction_test = np.argmax(y_predicted_test, axis=1)  # Trabajamos con clasificación multicategoría, no necesario para los bosques aleatorios
+        
+        acc_score_validation = accuracy_score(model.y_validation, y_final_prediction_validation)
+        print("Global accuracy score (validation) = "+str(round(acc_score_validation*100,2))+" [%]")
+        F1_score_validation = f1_score(model.y_validation, y_final_prediction_validation, average='macro')    # revisar las opciones de average
+        print("Global F1 score (validation) = "+str(round(F1_score_validation*100,2))+" [%]")
 
-    else:
+    else: # random forest, xgboost
         y_final_prediction_train = y_predicted_train
-        y_final_prediction_validation = y_predicted_validation
         y_final_prediction_test = y_predicted_test   # esta línea solo es necesaria para los bosques aleatorios y XGBoost
 
 
@@ -134,7 +138,7 @@ def tester(case_id_folder, model_id, create_superclasses, create_superclasses_CP
     cm = confusion_matrix(model.y_test, y_final_prediction_test, labels=all_classes)
 
     # Graficar la matriz de confusión
-    confusion_matrix_test_path = os.path.join(case_id_folder, "confusion_matrix_test_"+run_index+".png")
+    confusion_matrix_test_path = os.path.join(case_id_folder, "confusion_matrix_test_"+str(run_index)+".png")
 
     plt.figure(figsize=(10,7))
     sns.heatmap(cm, annot=True, fmt='d', cmap='Blues', xticklabels=class_names_total, yticklabels=class_names_total)
@@ -149,10 +153,7 @@ def tester(case_id_folder, model_id, create_superclasses, create_superclasses_CP
 
     # Accuracy
     acc_score_train = accuracy_score(model.y_train, y_final_prediction_train)
-    print("Global accuracy score (train) = "+str(round(acc_score_train*100,2))+" [%]")
-    
-    acc_score_validation = accuracy_score(model.y_validation, y_final_prediction_validation)
-    print("Global accuracy score (validation) = "+str(round(acc_score_validation*100,2))+" [%]")
+    print("Global accuracy score (train) = "+str(round(acc_score_train*100,2))+" [%]")    
     
     acc_score_test = accuracy_score(model.y_test, y_final_prediction_test)
     print("Global accuracy score (test) = "+str(round(acc_score_test*100,2))+" [%]")
@@ -161,9 +162,6 @@ def tester(case_id_folder, model_id, create_superclasses, create_superclasses_CP
     F1_score_train = f1_score(model.y_train, y_final_prediction_train, average='macro')    # revisar las opciones de average
     print("Global F1 score (train) = "+str(round(F1_score_train*100,2))+" [%]")
     
-    F1_score_validation = f1_score(model.y_validation, y_final_prediction_validation, average='macro')    # revisar las opciones de average
-    print("Global F1 score (validation) = "+str(round(F1_score_validation*100,2))+" [%]")
-    
     F1_score_test = f1_score(model.y_test, y_final_prediction_test, average='macro')    # revisar las opciones de average
     print("Global F1 score (test) = "+str(round(F1_score_test*100,2))+" [%]")
 
@@ -172,12 +170,13 @@ def tester(case_id_folder, model_id, create_superclasses, create_superclasses_CP
     # print("Global recall score = "+str(round(recall_score_global*100,2))+" [%]")
 
     # Save to a file
-    clasification_global_report_path = os.path.join(case_id_folder, "clasification_global_report_"+run_index+".txt")
+    clasification_global_report_path = os.path.join(case_id_folder, "clasification_global_report_"+str(run_index)+".txt")
     with open(clasification_global_report_path, "w") as f:
         f.write(f"Global F1 Score (train): {F1_score_train:.4f}\n")
         f.write(f"Global accuracy score (train): {acc_score_train:.4f}\n")
-        f.write(f"Global F1 Score (validation): {F1_score_validation:.4f}\n")
-        f.write(f"Global accuracy score (validation): {acc_score_validation:.4f}\n")
+        if model_id in [ML_Model.ESANN, ML_Model.CAPTURE24, ML_Model.XGBOOST]:
+            f.write(f"Global F1 Score (validation): {F1_score_validation:.4f}\n")
+            f.write(f"Global accuracy score (validation): {acc_score_validation:.4f}\n")
         f.write(f"Global F1 Score (test): {F1_score_test:.4f}\n")
         f.write(f"Global accuracy score (test): {acc_score_test:.4f}\n")
         # f.write(f"Global recall score: {recall_score_global:.4f}\n")
@@ -198,6 +197,6 @@ def tester(case_id_folder, model_id, create_superclasses, create_superclasses_CP
     print(classification_per_class_report)        
 
     # Save per-class report to a file
-    clasification_per_class_report_path = os.path.join(case_id_folder, "clasification_per_class_report_"+run_index+".txt")
+    clasification_per_class_report_path = os.path.join(case_id_folder, "clasification_per_class_report_"+str(run_index)+".txt")
     with open(clasification_per_class_report_path, "w") as f:        
         f.write(classification_per_class_report)

wearablepermed_ml/trainer.py

@@ -5,9 +5,9 @@ import logging
 from enum import Enum
 
 import numpy as np
-from data import DataReader
-from models.model_generator import modelGenerator
-from basic_functions.address import *
+from wearablepermed_ml.data import DataReader
+from wearablepermed_ml.models.model_generator import modelGenerator
+from wearablepermed_ml.basic_functions.address import *
 
 import tensorflow as tf
 
@@ -22,9 +22,11 @@ from ray.air import session
 
 from ray.tune.tuner import TuneConfig
 
-from models import SiMuRModel_ESANN, SiMuRModel_CAPTURE24, SiMuRModel_RandomForest, SiMuRModel_XGBoost
+from wearablepermed_ml.models import SiMuRModel_ESANN, SiMuRModel_CAPTURE24, SiMuRModel_RandomForest, SiMuRModel_XGBoost
 from sklearn.metrics import accuracy_score
 
+from sklearn.model_selection import cross_val_score
+
 
 # Configuration of GPU
 gpus = tf.config.list_physical_devices('GPU')
@@ -574,11 +576,11 @@ def main(args):
                 # ------------------------------------------------------------------------------------------------------
                 # Espacio de búsqueda
                 search_space = {
-                    "n_estimators": tune.randint(50, 301),                                  # Número de árboles entre 50 y 300
-                    "max_depth": tune.choice([5, 10, 15, 20, None]),                        # Profundidad máxima del árbol
-                    "min_samples_split": tune.randint(2, 11),                               # Muestras mínimas para dividir un nodo
-                    "min_samples_leaf": tune.randint(1, 11),                                # Muestras mínimas por hoja
-                    "max_features": tune.choice([None, "sqrt", "log2"]),                     # Número de características por división
+                    "n_estimators": tune.randint(200, 301),                                  # Número de árboles entre 50 y 300
+                    "max_depth": tune.choice([5, 6, 8]),                        # Profundidad máxima del árbol
+                    "min_samples_split": tune.randint(8, 15),                               # Muestras mínimas para dividir un nodo
+                    "min_samples_leaf": tune.randint(6, 10),                                # Muestras mínimas por hoja
+                    "max_features": tune.choice(["sqrt", "log2"]),                     # Número de características por división
                     "random_state": tune.randint(0, 10000)
                 }
 
@@ -694,15 +696,15 @@ def main(args):
                 # ------------------------------------------------------------------------------------------------------
                 # Espacio de búsqueda para XGBoost
                 search_space_xgb = {
-                    "num_boost_round": tune.randint(50, 3001),             # Árboles (rondas) de boosting
-                    "max_depth": tune.randint(3, 11),                      # Profundidad máxima
-                    "learning_rate": tune.uniform(0.01, 0.3),              # Tasa de aprendizaje
-                    "subsample": tune.uniform(0.5, 1.0),                   # Fracción de muestras por árbol
-                    "colsample_bytree": tune.uniform(0.5, 1.0),            # Fracción de columnas por árbol
-                    "gamma": tune.uniform(0, 5),                           # Regularización mínima de pérdida
-                    "min_child_weight": tune.randint(1, 10),               # Peso mínimo de hijos
-                    "reg_alpha": tune.uniform(0, 1),                       # L1 regularization
-                    "reg_lambda": tune.uniform(0, 1),                       # L2 regularization
+                    "num_boost_round": tune.randint(200, 1000),             # Árboles (rondas) de boosting
+                    "max_depth": tune.randint(2, 5),                      # Profundidad máxima
+                    "learning_rate": tune.uniform(0.01, 0.07),              # Tasa de aprendizaje
+                    "subsample": tune.uniform(0.4, 0.8),                   # Fracción de muestras por árbol
+                    "colsample_bytree": tune.uniform(0.4, 0.8),            # Fracción de columnas por árbol
+                    "gamma": tune.uniform(0, 10),                           # Regularización mínima de pérdida
+                    "min_child_weight": tune.randint(5, 50),               # Peso mínimo de hijos
+                    "reg_alpha": tune.loguniform(0.1, 100),                       # L1 regularization
+                    "reg_lambda": tune.loguniform(0.1, 100),                       # L2 regularization
                     "random_state": tune.randint(0, 10000)
                 }
 

uniovi_simur_wearablepermed_ml-1.1.0.dist-info/RECORD

@@ -1,19 +0,0 @@
-uniovi_simur_wearablepermed_ml-1.1.0.dist-info/licenses/LICENSE.txt,sha256=MJSf2LY2uh50n0Y7vPzbMcIfTXiY_IvEp7dQMhSHBWo,1089
-wearablepermed_ml/__init__.py,sha256=Jl6IIVd6LQLVAQG-uKC7nnY2204vk_YrBMz3bC4JxvU,601
-wearablepermed_ml/run_trainer_and_tester_30_times.py,sha256=6k7Sl6FAFhdYTQgM1MigUNXo5kAtYGAtZlTvXFd7oT8,8690
-wearablepermed_ml/tester.py,sha256=1Ac2l8MypZJTf1xl866G4dllH7r60iDWTBNMHieW40I,4346
-wearablepermed_ml/trainer.py,sha256=zoglrCPToTyQBEDPZ39rbmglnvjQcHghH2qX1lrEFRc,43621
-wearablepermed_ml/basic_functions/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-wearablepermed_ml/basic_functions/address.py,sha256=VYAvZLqyd4DvX2bYUJI5fnepKPMaI4CY7uiaeCsNauM,664
-wearablepermed_ml/data/DataReader.py,sha256=Bd1AqSytx0AXx1S_C9osk9I205vXdoNYP6IZctDSOmw,18476
-wearablepermed_ml/data/__init__.py,sha256=V0piawbJbkBykxziMrDr0W8KloG6TUyWhj_QDhlZDqw,34
-wearablepermed_ml/models/SiMuR_Model.py,sha256=WXAkvFhOhLkzgCjiwkmT4X9693-zlolnhsBepNCXarc,30346
-wearablepermed_ml/models/__init__.py,sha256=HkTS7w1Mpj0TP7ACtgcKv2gZEVYDNqAectiwGBI_Mmo,108
-wearablepermed_ml/models/model_generator.py,sha256=r23ahAE-otf0OGt_keD77dIb2-x9KN2ZP_e9ZhGEukk,1852
-wearablepermed_ml/testing/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-wearablepermed_ml/testing/testing.py,sha256=Uiq8AolTgbR2wcV12pUygSv3T8Y38hdpVC2yKoz04Ks,9927
-uniovi_simur_wearablepermed_ml-1.1.0.dist-info/METADATA,sha256=-g_EosjeCgXKo6h132EKu177_cDmv7q1d-cnwvNJVS0,15626
-uniovi_simur_wearablepermed_ml-1.1.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-uniovi_simur_wearablepermed_ml-1.1.0.dist-info/entry_points.txt,sha256=BW3Dw-fW_hquNHbK1q7pUNgIGwRDMPHqs9znQE6SECM,96
-uniovi_simur_wearablepermed_ml-1.1.0.dist-info/top_level.txt,sha256=PK7Cm_vvudFpRS-vGYiQDsZLkFZ6mdaocJqSPXHMP4c,18
-uniovi_simur_wearablepermed_ml-1.1.0.dist-info/RECORD,,