Update app.py

app.py

@@ -3,54 +3,29 @@ from definitions import *
 st.set_option('deprecation.showPyplotGlobalUse', False)
 st.sidebar.subheader("请选择模型参数:sunglasses:")
 
-num_leaves = st.sidebar.slider(label = 'num_leaves', min_value = 4.0,
-                          max_value = 16.0 ,
-                          value = 10.0,
-                          step = 0.1)
+num_leaves = st.sidebar.slider(label = 'num_leaves', min_value = 4,
+                          max_value = 200 ,
+                          value = 31,
+                          step = 1)
 
-max_depth = st.sidebar.slider(label = 'max_depth',  min_value = 8,
+max_depth = st.sidebar.slider(label = 'max_depth',  min_value = -1,
                           max_value = 15,
-                          value = 10,
+                          value = -1,
                           step = 1)
                           
 min_data_in_leaf = st.sidebar.slider(label = 'min_data_in_leaf',  min_value = 8,
-                          max_value = 15,
-                          value = 10,
+                          max_value = 55,
+                          value = 20,
                           step = 1)
 
 feature_fraction = st.sidebar.slider(label = 'feature_fraction', min_value = 0.0,
                           max_value = 1.0 ,
-                          value = 0.3,
-                          step = 0.1)
-
-lambda_l1 = st.sidebar.slider(label = 'lambda_l1', min_value = 0.000,
-                          max_value = 1.000 ,
-                          value = 0.500,
-                          step = 0.001)
-   
-lambda_l2 = st.sidebar.slider(label = 'lambda_l2', min_value = 1,
-                          max_value = 72,
-                          value = 36,
-                          step = 1)
-
-min_split_gain = st.sidebar.slider(label = 'min_split_gain', min_value = 6,
-                          max_value = 289 ,
-                          value = 144,
-                          step = 1)
-
-top_rate = st.sidebar.slider(label = 'top_rate', min_value = 0.0,
-                          max_value = 1.0 ,
-                          value = 0.3,
-                          step = 0.1)
-
-other_rate = st.sidebar.slider(label = 'other_rate', min_value = 0.0,
-                          max_value = 1.0 ,
-                          value = 0.3,
+                          value = 0.8,
                           step = 0.1)
                           
 min_data_per_group = st.sidebar.slider(label = 'min_data_per_group', min_value = 6,
                           max_value = 289 ,
-                          value = 32,
+                          value = 100,
                           step = 1)
 
 max_cat_threshold = st.sidebar.slider(label = 'max_cat_threshold', min_value = 6,
@@ -58,21 +33,15 @@ max_cat_threshold = st.sidebar.slider(label = 'max_cat_threshold', min_value = 6
                           value = 32,
                           step = 1)
                           
-learning_rate = st.sidebar.slider(label = 'learning_rate', min_value = 8.0,
-                          max_value = 15.0,
-                          value = 10.5,
-                          step = 0.1)
+learning_rate = st.sidebar.slider(label = 'learning_rate', min_value = 0.0,
+                          max_value = 1.00,
+                          value = 0.05,
+                          step = 0.01)
 
 num_leaves = st.sidebar.slider(label = 'num_leaves',  min_value = 6,
                           max_value = 289 ,
                           value = 31,
                           step = 1)
-                          
-min_gain_to_split  = st.sidebar.slider(label = 'min_gain_to_split', min_value = 0.0,
-                          max_value = 15.0,
-                          value = 0.0,
-                          step = 0.1)
-
 
 max_bin = st.sidebar.slider(label = 'max_bin', min_value = 6,
                           max_value = 289 ,
@@ -80,8 +49,8 @@ max_bin = st.sidebar.slider(label = 'max_bin', min_value = 6,
                           step = 1)
 
 num_iterations = st.sidebar.slider(label = 'num_iterations', min_value = 8,
-                          max_value = 15,
-                          value = 10,
+                          max_value = 289,
+                          value = 100,
                           step = 1)
                                  
 st.title('LightGBM-parameter-tuning-with-streamlit')
@@ -103,11 +72,9 @@ lgb_eval = lgb.Dataset(X_test, y_test, reference=lgb_train)
 params = {'num_leaves': num_leaves, 'max_depth': max_depth,
             'min_data_in_leaf': min_data_in_leaf, 
             'feature_fraction': feature_fraction,
-            'lambda_l1': lambda_l1, 'lambda_l2': lambda_l2,
-            'min_split_gain': min_split_gain, 'top_rate': top_rate,
-            'other_rate': other_rate, 'min_data_per_group': min_data_per_group, 
+            'min_data_per_group': min_data_per_group, 
             'max_cat_threshold': max_cat_threshold,
-            'learning_rate':learning_rate,'num_leaves':num_leaves,'min_gain_to_split':min_gain_to_split,
+            'learning_rate':learning_rate,'num_leaves':num_leaves,
             'max_bin':max_bin,'num_iterations':num_iterations
             }
 
@@ -118,11 +85,8 @@ probs = gbm.predict(X_test, num_iteration=gbm.best_iteration)  # 输出的是概
 fpr, tpr, thresholds = roc_curve(y_test, probs)
 st.write('------------------------------------')
 st.write('Confusion Matrix:')
-
 st.write(confusion_matrix(y_test, np.where(probs > 0.5, 1, 0)))
 
-
-
 st.write('------------------------------------')
 st.write('Classification Report:')
 report = classification_report(y_test, np.where(probs > 0.5, 1, 0), output_dict=True)