聚烯烃添加特征重要度图

lib/dataread.py

@@ -553,6 +553,61 @@ def featureAnalysis(df, dataset, y):
     #         plt.close()
 
 
+# 计算特征重要度
+def feature_importance(X_train, y_train):
+    import lightgbm as lgb
+    import matplotlib.pyplot as plt
+    from xgboost import XGBRegressor
+    from xgboost import plot_importance
+
+    # 特征选择
+    plt.figure(figsize=(2, 15))
+
+    plt.rcParams['font.sans-serif'] = ['SimHei']
+
+    temp = XGBRegressor()
+    temp.fit(X_train, y_train)
+
+    ax = plot_importance(temp)
+    fig = ax.figure
+    fig.set_size_inches(8, 7)
+
+    # 修改图的标题，添加模型名称
+    title = '特征重要度1'  # 替换为你的模型名称
+    ax.set_title(title)
+
+    # plt.show()
+    # 保存图片
+    plt.savefig(os.path.join(config.dataset, '特征重要度1.png'))
+    plt.close()
+    config.logger.info('特征重要度1.png 已保存')
+
+    # 创建一个 LGBMRegressor 对象并训练模型
+    regressor = lgb.LGBMRegressor()
+    regressor.fit(X_train, y_train)
+
+    # 设置图形大小（可选）
+    plt.figure(figsize=(30, 40))
+    # 使用 plot_importance 函数来绘制特征重要性
+    # 注意：在一些版本的 LightGBM 中，你可以直接传入模型对象
+    ax = importance_plot = lgb.plot_importance(
+        regressor, importance_type='gain')  # 或者 'split'
+    # 设置标题和字体大小
+    ax.set_title('Feature Importance - LGBMRegressor', fontsize=12)
+    for item in ([ax.title, ax.xaxis.label, ax.yaxis.label] +
+                 ax.get_xticklabels() + ax.get_yticklabels()):
+        item.set_fontsize(9)  # 设置 x 轴、y 轴标签以及刻度标签的字体大小
+    # 修改图的标题，添加模型名称
+    # title = 'Feature Importance - LGBMRegressor '  # 替换为你的模型名称
+    # ax.set_title(title)
+    # 保存图片
+    plt.savefig(os.path.join(config.dataset, '特征重要度2.png'))
+
+    # 显示图形
+    plt.close()
+    config.logger.info('特征重要度2.png 已保存')
+
+
 def corr_feature(df):
     # 重新命名列名，列名排序,y在第一个
     df.reindex(['y'] + sorted(df.columns.difference(['y'])))
@@ -1102,6 +1157,7 @@ def datachuli_juxiting(df_zhibiaoshuju, df_zhibiaoliebiao, datecol='date', end_t
 
     calculate_correlation(df=df)
     featureAnalysis(df, dataset=dataset, y='y')
+
     return df
 
 

models/nerulforcastmodels.py

@@ -349,6 +349,11 @@ def ex_Model_Juxiting(df, horizon, input_size, train_steps, val_check_steps, ear
     config.logger.info("\nTesting set head:")
     config.logger.info(df_test.head())
 
+    # 特征重要度
+    X_train = df_train.drop(columns=['y'])
+    y_train = df_train['y']
+    feature_importance(X_train=X_train, y_train=y_train)
+
     models = [
         NHITS(h=horizon, input_size=input_size, max_steps=train_steps, val_check_steps=val_check_steps,
               scaler_type='standard', activation='ReLU', early_stop_patience_steps=early_stop_patience_steps),
@@ -396,7 +401,8 @@ def ex_Model_Juxiting(df, horizon, input_size, train_steps, val_check_steps, ear
 
         # VanillaTransformer(h=horizon, input_size=input_size, max_steps=train_steps, val_check_steps=val_check_steps, scaler_type='standard', ),  //报错了
         # Autoformer(h=horizon, input_size=input_size, max_steps=train_steps, val_check_steps=val_check_steps, scaler_type='standard', ), //报错了
-        NBEATS(h=horizon, input_size=input_size, max_steps=train_steps, val_check_steps=val_check_steps, scaler_type='standard', ),
+        NBEATS(h=horizon, input_size=input_size, max_steps=train_steps,
+               val_check_steps=val_check_steps, scaler_type='standard', ),
         # NBEATSx (h=horizon, input_size=input_size, max_steps=train_steps, val_check_steps=val_check_steps, scaler_type='standard',activation='ReLU', ),   //报错
         # HINT(h=horizon),
 
@@ -462,8 +468,8 @@ def ex_Model_Juxiting(df, horizon, input_size, train_steps, val_check_steps, ear
     df_predict['created_dt'] = end_time
 
     # 预测结果保留整数（先四舍五入再转换为整数类型）
-    df_predict = df_predict.round().astype({col: 'int' for col in df_predict.columns if col not in ['ds', 'created_dt']})
+    df_predict = df_predict.round().astype(
-
+        {col: 'int' for col in df_predict.columns if col not in ['ds', 'created_dt']})
 
     # 保存预测值
     df_predict.to_csv(os.path.join(config.dataset, "predict.csv"), index=False)
@@ -2055,7 +2061,6 @@ def model_losss_juxiting(sqlitedb, end_time, is_fivemodels):
     except ValueError as e:
         print(e)
 
-
     def add_upper_lower_bound(row):
         print(row['columns'])
         print(type(row['columns']))
@@ -3343,7 +3348,8 @@ def pp_export_pdf(num_indicators=475, num_models=21, num_dayindicator=202, input
         stime = df3['ds'].iloc[0]
         etime = df3['ds'].iloc[-1]
         # 添加偏差率表格
-        fivemodels = '、'.join(eval_df['模型(Model)'].values[:5])  # 字符串形式，后面写入字符串使用
+        fivemodels = '、'.join(
+            eval_df['模型(Model)'].values[:5])  # 字符串形式，后面写入字符串使用
         content.append(Graphs.draw_text(
             f'预测使用了{num_models}个模型进行训练，使用评估结果MAE前五的模型分别是 {fivemodels} ，模型上一预测区间 {stime} -- {etime}的偏差率(%)分别是：'))
         # # 添加偏差率表格
@@ -3541,34 +3547,35 @@ def pp_export_pdf(num_indicators=475, num_models=21, num_dayindicator=202, input
 def pp_bdwd_png(global_config):
     best_bdwd_price = find_best_models(
         date=global_config['end_time'], global_config=global_config)
-    y_hat_yuedu = pd.DataFrame(best_bdwd_price).T[['date', 'predictresult']][-4:]
+    y_hat_yuedu = pd.DataFrame(
+        best_bdwd_price).T[['date', 'predictresult']][-4:]
     y_hat_yuedu['ds'] = pd.to_datetime(y_hat_yuedu['date'])
     # 绘制PP期货月度预测结果的图表
     plot_pp_predict_result(y_hat_yuedu, global_config)
 
-    y_hat_zhoudu = pd.DataFrame(best_bdwd_price).T[['date', 'predictresult']][2:4]
+    y_hat_zhoudu = pd.DataFrame(
+        best_bdwd_price).T[['date', 'predictresult']][2:4]
     y_hat_zhoudu['ds'] = pd.to_datetime(y_hat_zhoudu['date'])
-    y_hat_zhoudu.drop(columns=['date'],inplace=True)
+    y_hat_zhoudu.drop(columns=['date'], inplace=True)
     print(y_hat_zhoudu)
     # 获取本周最佳模型的五日预测价格
     five_days_predict_price = pd.read_csv('juxitingdataset/predict.csv')
     week_price_modelname = best_bdwd_price['week_price']['model_name']
-    five_days_predict_price = five_days_predict_price[['ds',week_price_modelname]]
+    five_days_predict_price = five_days_predict_price[[
-    five_days_predict_price['ds'] = pd.to_datetime(five_days_predict_price['ds'])
+        'ds', week_price_modelname]]
-    five_days_predict_price.rename(columns={week_price_modelname:'predictresult'},inplace=True)
+    five_days_predict_price['ds'] = pd.to_datetime(
+        five_days_predict_price['ds'])
+    five_days_predict_price.rename(
+        columns={week_price_modelname: 'predictresult'}, inplace=True)
     # 设置索引 次日 次二日 次三日 次四日  次五日
     index_labels = ["次日", "次二日", "次三日", "次四日", "次五日"]
     five_days_predict_price.index = index_labels
-    y_hat_riduzhoudu = pd.concat([y_hat_zhoudu, five_days_predict_price], axis=0)
+    y_hat_riduzhoudu = pd.concat(
+        [y_hat_zhoudu, five_days_predict_price], axis=0)
     y_hat_riduzhoudu = y_hat_riduzhoudu.sort_values(by='ds')
     print(y_hat_riduzhoudu)
     # 绘制PP期货日度周度预测结果的图表
-    plot_pp_predict_result(y_hat_riduzhoudu, global_config,'zhoudu')
+    plot_pp_predict_result(y_hat_riduzhoudu, global_config, 'zhoudu')
-
-
-
-
-
 
 
 def pp_export_pdf_v1(num_indicators=475, num_models=21, num_dayindicator=202, inputsize=5, dataset='dataset', time='2024-07-30', reportname='report.pdf'):