feature_importance

main.py

@@ -55,7 +55,7 @@ class LightGBM(object):
         ]
         self.split_c = 0.98
         self.yc = 0.8
-        self.model = "lightgbm_tag_train.bin"
+        self.model = "lightgbm_tag_train_01.bin"
 
     def generate_x_data(self):
         """
@@ -110,14 +110,14 @@ class LightGBM(object):
             "objective": "binary",  # 指定二分类任务
             "metric": "binary_logloss",  # 评估指标为二分类的log损失
             "num_leaves": 31,  # 叶子节点数
-            "learning_rate": 0.05,  # 学习率
+            "learning_rate": 0.01,  # 学习率
             "bagging_fraction": 0.9,  # 建树的样本采样比例
             "feature_fraction": 0.8,  # 建树的特征选择比例
             "bagging_freq": 5,  # k 意味着每 k 次迭代执行bagging
             "num_threads": 4,  # 线程数量
         }
         # 训练模型
-        num_round = 100
+        num_round = 500
         print("开始训练......")
         bst = lgb.train(params, train_data, num_round, valid_sets=[test_data])
         bst.save_model(self.model)
@@ -181,6 +181,6 @@ class LightGBM(object):
 
 if __name__ == "__main__":
     L = LightGBM()
-    # L.train_model()
+    L.train_model()
     # L.evaluate_model()
     L.feature_importance()