适配聚类模型的代码生成逻辑;

.idea/mongoSettings.xml

@@ -0,0 +1,20 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<project version="4">
+  <component name="MongoConfiguration">
+    <option name="serverConfigurations">
+      <list>
+        <ServerConfiguration>
+          <option name="label" value="local_mongo" />
+          <option name="serverUrls">
+            <list>
+              <option value="localhost:27017" />
+            </list>
+          </option>
+          <option name="sshTunnelingConfiguration">
+            <SshTunnelingConfiguration />
+          </option>
+        </ServerConfiguration>
+      </list>
+    </option>
+  </component>
+</project>

Engines/engine_urls.py

@@ -1,15 +1,17 @@
 from django.urls import path
+
 from .engine_view import *
-from utils.views2urls import views2urls
 
 app_name = 'engine'
 
-
 # urlpatterns = views2urls("Engines.engine_view")
 urlpatterns = [
     # 校验清洗规则
     path('check_clean_condition', check_clean_condition),
     # 清洗完成的数据保存
-    path('save_clean_data', save_clean_data)
+    path('save_clean_data', save_clean_data),
+    # 建模特征列检测
+    path('check_mining_condition', check_mining_condition),
+    # 运行生成的建模代码,并生成报告
+    path('run_mining_code', run_mining_code)
 ]
-

Engines/engine_view.py

@@ -1,5 +1,6 @@
 import json
 import traceback
+
 import pandas as pd
 from django.http import JsonResponse
 from django.views.decorators.http import require_http_methods
@@ -47,7 +48,7 @@ def save_clean_data(request):
         conditions = post_body.pop("conditions", {})
         new_data = clean_engine.check_clean_condition(user_name, dataset_name, conditions)
         if isinstance(new_data, pd.core.frame.DataFrame):
-            res = clean_engine.save_clean_data(user_name,dataset_name,new_data)
+            res = clean_engine.save_clean_data(user_name, dataset_name, new_data)
             if res:
                 msg = "保存成功"
     except Exception as e:
@@ -56,3 +57,43 @@ def save_clean_data(request):
         traceback.print_exc()
 
     return JsonResponse({"code": code, "msg": msg, "data": data})
+
+
+@require_http_methods(['POST'])
+def check_mining_condition(request):
+    """
+    检测建模条件， 对特征列类型的合理性进行校验
+    :param request:
+    :return:
+    """
+    data, code, msg = None, 200, None
+    try:
+        post_body = json.loads(request.body)
+        postBody = json.loads(request.body)
+        username = postBody.pop('username')
+        postBody = postBody.get('data')
+        name = postBody.get('name')
+        dataset_name = postBody.get('dataset_name')
+        features = postBody.get('features')
+        target = postBody.get('target')
+        model_type = postBody.get('model_type')
+        model_name = postBody.get('models')
+        evaluate_methods = postBody.get("metrics")
+    except Exception as e:
+        msg = str(e)
+        coce = 500
+
+    return JsonResponse({"code": code, "msg": msg, "data": data})
+
+
+@require_http_methods(['POST'])
+def run_mining_code(request):
+    data, code, msg = None, 200, None
+    try:
+        post_body = json.loads(request.body)
+        dataset_name = post_body.pop("dataset", "")
+        user_name = post_body.pop("user_name", "")
+        conditions = post_body.pop("conditions", {})
+    except Exception as e:
+        msg, code = str(e), 500
+    return JsonResponse({"code": code, "msg": msg, "data": data})

ModelSelection/dataset_process_model.py

@@ -103,7 +103,7 @@ class DatasetProcess:
             res = model.find_one(query)
             return res['data']
         except Exception as e:
-            traceback.print_exc()
+            print(str(e))
             raise e
 
     def get_dataset_cols(self, dataset_name):

ModelSelection/model_process_model.py

@@ -1,7 +1,5 @@
 import os
 
-
-
 """
 流程：   
 ` 1、从monodb读取用户上传的数据集（已完成）`       
@@ -89,7 +87,7 @@ class SetModel():
         self.joint_code('ImportPackages.py')
         # 拼接模型需要的库
         for model in self.model_name:
-            self.generate += MODEL_DICT[self.model_type][model]+"\n"
+            self.generate += MODEL_DICT[self.model_type][model] + "\n"
 
         # 拼接变量
         self.generate += """
@@ -102,13 +100,17 @@ TARGET='{}'
         self.generate += """
 MODEL = [{}]
         """.format(", ".join(sklearn_models))
-        # 拼接主函数
-        self.joint_code('Main.py')
-        # 拼接分类/回归的必要评估方法
+
+        # 拼接分类/回归/聚类的主函数与必要评估方法
         if self.model_type == "分类":
-            self.joint_code("classifier_evaluation.py")
+            self.joint_code('main_supervisied.py')
+            self.joint_code("evaluation_classifier.py")
         elif self.model_type == "回归":
-            self.joint_code("regressor_evaluation.py")
+            self.joint_code('main_supervisied.py')
+            self.joint_code("evaluation_regressor.py")
+        elif self.model_type == "聚类":
+            self.joint_code("main_unsupervised.py")
+            self.joint_code("evaluation_cluster.py")
 
         # 拼接用户自选的可视化的评估方法
         if len(self.evaluate_methods) != 0:

ModelSelection/model_selection_views.py

@@ -262,3 +262,8 @@ def get_methods(request):
         else:
             res[method] = sub_method
     return JsonResponse({"data": res}, status=200)
+
+
+@require_http_methods(['POST'])
+def plot_graph(request):
+    pass

ModelSelection/urls.py

@@ -14,6 +14,7 @@ urlpatterns = [
     path('generate_code', generate_code),  # 生成代码
     path('export_code', export_code),  # 导出代码文件
     path('del_dataset', del_dataset),  # 删除数据集
+    path('plot_graph', plot_graph)  # 绘制图形
 
 ]
 

code_templates/evaluation_cluster.py

@@ -0,0 +1,27 @@
+# 模型性能评估
+def model_performance_evaluation(model_name, X, labels, spend_time):
+    from sklearn.metrics import calinski_harabaz_score
+    from sklearn.metrics import silhouette_score
+    sc = silhouette_score(X,labels)
+    chs = calinski_harabaz_score(X,labels)
+    print(model_name, "| 轮廓系数: %.4f" % sc)
+    print(model_name, "|  Calinski-Harabaz index: %.4f" % chs)
+    print(model_name, "| 训练时长(秒): %.4f" % spend_time)
+    return sc,chs
+
+
+# 使用模型预测并评估
+evaluation_dimensions = ["模型名称", "轮廓系数", "Calinski-Harabaz-index", "训练时长(秒)"]
+compare_result = {dimension: [] for dimension in evaluation_dimensions}
+
+for spend_time, fit_model in fit_models.items():
+    print("====================")
+    model_name = str(fit_model).split("(")[0]
+    labels = fit_model.labels_
+    sc, chs = model_performance_evaluation(model_name, X, labels, spend_time)
+    compare_result["模型名称"].append(model_name)
+    compare_result["轮廓系数"].append(sc)
+    compare_result["Calinski-Harabaz-index"].append(chs)
+    compare_result["训练时长(秒)"].append(spend_time)
+res_df = pd.DataFrame(compare_result)
+print(res_df)

code_templates/main_supervisied.py

@@ -1,4 +1,3 @@
-
 # 读取数据
 if FILE_PATH.split('.')[-1]=='xls' or FILE_PATH.split('.')[-1]=='xlsx':
     DF=pd.read_excel(FILE_PATH)

code_templates/main_unsupervised.py

@@ -0,0 +1,18 @@
+# 读取数据
+if FILE_PATH.split('.')[-1]=='xls' or FILE_PATH.split('.')[-1]=='xlsx':
+    DF=pd.read_excel(FILE_PATH)
+else:
+    DF=pd.read_csv(FILE_PATH)
+X = DF[FEATURES]
+# 模型训练,key为模型训练的时间,value为训练好的模型
+fit_models = {}
+for model in MODEL:
+    start_time = time.time()
+    fit_model= model.fit(X)
+    end_time = time.time()
+    spend_time = end_time - start_time
+    fit_models[spend_time] = fit_model
+
+
+
+

utils/MODEL_DICT.py

@@ -20,7 +20,12 @@ MODEL_DICT = {
         '支持向量机': 'from sklearn.svm import SVR',
         '神经网络': 'from sklearn.neural_network import MLPRegressor',
         'Adaboost': 'from sklearn.ensemble import AdaBoostRegressor',
-        '随机森林':'from sklearn.ensemble import RandomForestRegressor'
+        '随机森林': 'from sklearn.ensemble import RandomForestRegressor'
+    },
+    '聚类': {
+        'K_means': 'from sklearn.cluster import KMeans',
+        'MiniBatchKMeans': 'from sklearn.cluster import MiniBatchKMeans',
+        'BIRCH': 'from sklearn.cluster import Birch'
     },
     'ROC曲线': 'plot_ROC_curve.py',
     '混淆矩阵': 'plot_confusion_matrix.py'
@@ -35,7 +40,7 @@ METRICS_DICT = {
         "误差平方和": "",
         "决定系数(R²)": ""
     },
-    "聚类":{}
+    "聚类": {}
 }
 
 CLEAN_DICT = {