Evaluator save and returns to execute predict

MachineLearning.py

@@ -61,7 +61,7 @@ def _roc_values(testing, evaluation):
     print("\nRoc Score")
     print(auc_roc)
 
-    return auc_roc
+    return round(auc_roc,2)
 
 
 def _classification_report(test, evaluation):

config.py

@@ -7,3 +7,4 @@
 
 DATA_PATH = BASE_PATH + 'data/'
 MODEL_PATH = BASE_PATH + 'model/'
+EVALUATE_PATH = BASE_PATH + 'evaluate/'

evaluate/results.json

@@ -0,0 +1 @@
+{"results": [{"model": "cardio_dt", "inductor": "Decision Tree", "acc": 0.63, "sen": 0.63, "spe": 0.62, "roc": 0.63}, {"model": "cardio_lr", "inductor": "Logistic Regression", "acc": 0.71, "sen": 0.74, "spe": 0.67, "roc": 0.71}, {"model": "cardio_rf", "inductor": "Random Forest", "acc": 0.72, "sen": 0.73, "spe": 0.7, "roc": 0.72}, {"model": "cardio_svm_poly", "inductor": "SVM Kernel Poly", "acc": 0.72, "sen": 0.73, "spe": 0.7, "roc": 0.72}]}

trabalho_final/evaluator.py

@@ -0,0 +1,17 @@
+import json
+import config
+
+def evaluate_best_model():
+    filename = config.EVALUATE_PATH + 'results.json'
+
+    best_model = 0
+    with open(filename) as json_file:
+        data = json.load(json_file)
+        for p in data['results']:
+            sum = p['acc']+p['sen']+p['spe']+p['roc']
+            if sum > best_model:
+                best_model = sum
+                sel_model = p['model']
+                name_model = p['inductor']
+
+    return sel_model, name_model

trabalho_final/indutores.py

@@ -1,60 +1,129 @@
 from trabalho_final.preprocessing import get_data
 from MachineLearning import MachineLearning, accuracy, sensitivity, specificity
+import json
+import config
 
 machine_learning = MachineLearning()
 attributes, classes = get_data()
 
+json_data = {}
+json_data['results'] = []
+
 # Decision Tree
 print("\nGerando modelo de Decision Tree")
 response, auc_roc, matrix, report = machine_learning.generate_decision_tree(attributes, classes, 'cardio_dt')
 print(response)
 print("\nAcurácia")
-print(accuracy(matrix))
+acc = accuracy(matrix)
+print(acc)
 print("\nSensibilidade")
-print(sensitivity(matrix))
+sen = sensitivity(matrix)
+print(sen)
 print("\nEspecificidade")
-print(specificity(matrix))
+spe = specificity(matrix)
+print(spe)
+
+json_data['results'].append({
+    'model': 'cardio_dt',
+    'inductor': 'Decision Tree',
+    'acc': acc,
+    'sen': sen,
+    'spe': spe,
+    'roc': auc_roc,
+})
 
 # Logistic Regression
 print("\nGerando modelo de Logistic Regression")
 response, auc_roc, matrix, report = machine_learning.generate_logistic_regression(attributes, classes, 'cardio_lr')
 print(response)
 print("\nAcurácia")
-print(accuracy(matrix))
+acc = accuracy(matrix)
+print(acc)
 print("\nSensibilidade")
-print(sensitivity(matrix))
+sen = sensitivity(matrix)
+print(sen)
 print("\nEspecificidade")
-print(specificity(matrix))
+spe = specificity(matrix)
+print(spe)
+
+json_data['results'].append({
+    'model': 'cardio_lr',
+    'inductor': 'Logistic Regression',
+    'acc': acc,
+    'sen': sen,
+    'spe': spe,
+    'roc': auc_roc,
+})
 
 # Random Forest
 print("\nGerando modelo de Random Forest")
 response, auc_roc, matrix, report = machine_learning.generate_random_forest(attributes, classes, 'cardio_rf')
 print(response)
 print("\nAcurácia")
-print(accuracy(matrix))
+acc = accuracy(matrix)
+print(acc)
 print("\nSensibilidade")
-print(sensitivity(matrix))
+sen = sensitivity(matrix)
+print(sen)
 print("\nEspecificidade")
-print(specificity(matrix))
+spe = specificity(matrix)
+print(spe)
 
-# SVM kernel linear
-print("\nGerando modelo de SVM Kernel Linear")
-response, auc_roc, matrix, report = machine_learning.generate_svm(attributes, classes, 'cardio_svm_linear')
-print(response)
-print("\nAcurácia")
-print(accuracy(matrix))
-print("\nSensibilidade")
-print(sensitivity(matrix))
-print("\nEspecificidade")
-print(specificity(matrix))
+json_data['results'].append({
+    'model': 'cardio_rf',
+    'inductor': 'Random Forest',
+    'acc': acc,
+    'sen': sen,
+    'spe': spe,
+    'roc': auc_roc,
+})
 
-# SVM kernel poly
-print("\nGerando modelo de SVM Kernel Poly")
-response, auc_roc, matrix, report = machine_learning.generate_svm(attributes, classes, 'cardio_svm_poly', kernel='poly')
-print(response)
-print("\nAcurácia")
-print(accuracy(matrix))
-print("\nSensibilidade")
-print(sensitivity(matrix))
-print("\nEspecificidade")
-print(specificity(matrix))
+# # SVM kernel linear
+# print("\nGerando modelo de SVM Kernel Linear")
+# response, auc_roc, matrix, report = machine_learning.generate_svm(attributes, classes, 'cardio_svm_linear')
+# print(response)
+# print("\nAcurácia")
+# acc = accuracy(matrix)
+# print(acc)
+# print("\nSensibilidade")
+# sen = sensitivity(matrix)
+# print(sen)
+# print("\nEspecificidade")
+# spe = specificity(matrix)
+# print(spe)
+#
+# json_data['results'].append({
+#     'model': 'cardio_svm_linear',
+#     'inductor': 'SVM Kernel Linear',
+#     'acc': acc,
+#     'sen': sen,
+#     'spe': spe,
+#     'roc': auc_roc,
+# })
+#
+# # SVM kernel poly
+# print("\nGerando modelo de SVM Kernel Poly")
+# response, auc_roc, matrix, report = machine_learning.generate_svm(attributes, classes, 'cardio_svm_poly', kernel='poly')
+# print(response)
+# print("\nAcurácia")
+# acc = accuracy(matrix)
+# print(acc)
+# print("\nSensibilidade")
+# sen = sensitivity(matrix)
+# print(sen)
+# print("\nEspecificidade")
+# spe = specificity(matrix)
+# print(spe)
+
+json_data['results'].append({
+    'model': 'cardio_svm_poly',
+    'inductor': 'SVM Kernel Poly',
+    'acc': acc,
+    'sen': sen,
+    'spe': spe,
+    'roc': auc_roc,
+})
+
+filename = config.EVALUATE_PATH + 'results.json'
+with open(filename, 'w') as outfile:
+    json.dump(json_data, outfile)

trabalho_final/inferencia.py

@@ -1,4 +1,7 @@
 from MachineLearning import execute_model
+from trabalho_final.evaluator import evaluate_best_model
+
+model_exec, name_model = evaluate_best_model()
 
 params = [
     [22584, 2, 178, 95.0, 130, 90, 3, 3, 0, 0, 1],  # Y
@@ -9,27 +12,7 @@
     [8622, 2, 188, 90.0, 130, 90, 1, 1, 0, 0, 1],  # ?
 ]
 
-# Decision Tree
-print("\nTestando modelo Decision Tree")
-response = execute_model('cardio_dt', params)
-print(response)
-
-# Logistic Regression
-print("\nTestando modelo Logistic Regression")
-response = execute_model('cardio_lr', params)
-print(response)
-
-# Random Forest
-print("\nTestando modelo Random Forest")
-response = execute_model('cardio_rf', params)
-print(response)
-
-# SVM kernel linear
-print("\nTestando modelo SVM Kernel Linear")
-response = execute_model('cardio_svm_linear', params)
-print(response)
-
-# SVM kernel poly
-print("\nTestando modelo SVM Kernel Poly")
-response = execute_model('cardio_svm_poly', params)
+# Execute predict with best model evaluate
+print("\nTestando modelo "+name_model)
+response = execute_model(model_exec, params)
 print(response)