提交近期修改, 主要针对价格分位和余票变化情况

data_loader.py

@@ -602,18 +602,28 @@ def plot_c12_trend(df, output_dir="."):
             (g.index == len(g) - 1)  # 终点
         ].drop_duplicates(subset=['update_hour'])
 
-        # 绘制点和线条
-        plt.plot(
+        # 绘制阶梯线：价格在相邻区间内保持不变，在下一个时间点发生跳变
+        plt.step(
             change_points['update_hour'],
             change_points['adult_total_price'],
-            marker='o',
+            where='post',
             color=colors[i % len(colors)],
             linestyle=linestyles[i % len(linestyles)],
-            linewidth=2, markersize=6,
-            markerfacecolor='white', markeredgewidth=2,
+            linewidth=2,
             label=f"Baggage {baggage_value}"
         )
-
+        # 单独绘制变化点，保留原来的圆点视觉效果
+        plt.plot(
+            change_points['update_hour'],
+            change_points['adult_total_price'],
+            linestyle='None',
+            marker='o',
+            color=colors[i % len(colors)],
+            markersize=6,
+            markerfacecolor='white',
+            markeredgewidth=2,
+        )
+        
         # 添加注释 (小时数, 价格)
         for _, row in change_points.iterrows():
             text = f"({row['hours_until_departure']}, {row['adult_total_price']})"
@@ -1190,7 +1200,7 @@ if __name__ == "__main__":
     os.makedirs(output_dir, exist_ok=True)
 
     # 加载热门航线数据
-    date_begin = "2026-04-01"
+    date_begin = "2026-04-21"
     date_end = datetime.today().strftime("%Y-%m-%d")
 
     flight_route_list = vj_flight_route_list_nothot[:]  # 热门 vj_flight_route_list_hot  冷门 vj_flight_route_list_nothot

data_preprocess.py

@@ -855,7 +855,7 @@ def preprocess_data_simple(df_input, is_train=False):
         df_input = df_input[~((df_input['is_filled'] == 1) & (_tail_filled == 1))]
     
     # 价格变化最小量阈值
-    price_change_amount_threshold = 1
+    price_change_amount_threshold = 0.001
     df_input['_raw_price_diff'] = df_input.groupby(['gid', 'baggage'], group_keys=False)['adult_total_price'].diff()
 
     # 计算价格变化量
@@ -905,6 +905,30 @@ def preprocess_data_simple(df_input, is_train=False):
         .add(1)
     )
     
+    # 第三步：段级余票变化（上一段终点余票 -> 当前段起点余票），首段为 n->n
+    _seg_keys = ['gid', 'baggage', 'price_change_segment']
+    _seg_seats = (
+        df_input.groupby(_seg_keys, as_index=False)['seats_remaining']
+        .agg(_seg_first_seats='first', _seg_last_seats='last')
+    )
+    _seg_seats['_prev_seg_last_seats'] = (
+        _seg_seats.groupby(['gid', 'baggage'], group_keys=False)['_seg_last_seats']
+        .shift(1)
+        .fillna(_seg_seats['_seg_first_seats'])
+    )
+    
+    _seg_seats['_seg_first_seats'] = pd.to_numeric(_seg_seats['_seg_first_seats'], errors='coerce').round().astype('Int64')
+    _seg_seats['_prev_seg_last_seats'] = pd.to_numeric(_seg_seats['_prev_seg_last_seats'], errors='coerce').round().astype('Int64')
+    _seg_seats['seats_remaining_transition'] = (
+        _seg_seats['_prev_seg_last_seats'].astype(str) + '->' + _seg_seats['_seg_first_seats'].astype(str)
+    )
+
+    df_input = df_input.merge(
+        _seg_seats[_seg_keys + ['seats_remaining_transition']],
+        on=_seg_keys,
+        how='left'
+    )
+
     # 可选：删除临时列
     # df_input = df_input.drop(columns=['price_change_segment'])
     df_input = df_input.drop(columns=['price_change_segment', '_raw_price_diff'])
@@ -929,7 +953,7 @@ def preprocess_data_simple(df_input, is_train=False):
 
     # 训练过程
     if is_train:
-        df_target = df_input[(df_input['hours_until_departure'] >= 72) & (df_input['hours_until_departure'] <= 360)].copy()   # 扩展至360小时（15天） 
+        df_target = df_input[(df_input['hours_until_departure'] >= 48) & (df_input['hours_until_departure'] <= 384)].copy()   # 扩展至360小时（15天） 
         df_target = df_target.sort_values(
             by=['gid', 'hours_until_departure'],
             ascending=[True, False]
@@ -940,7 +964,7 @@ def preprocess_data_simple(df_input, is_train=False):
         prev_amo = df_target.groupby('gid', group_keys=False)['price_change_amount'].shift(1)
         prev_dur = df_target.groupby('gid', group_keys=False)['price_duration_hours'].shift(1)
         prev_price = df_target.groupby('gid', group_keys=False)['adult_total_price'].shift(1)
-        prev_seats = df_target.groupby('gid', group_keys=False)['seats_remaining'].shift(1)
+        prev_seats_trans = df_target.groupby('gid', group_keys=False)['seats_remaining_transition'].shift(1)
 
         # 对于先升后降（先降后降）的分析
         seg_start_mask = df_target['price_duration_hours'].eq(1)   # 开始变价节点
@@ -963,7 +987,7 @@ def preprocess_data_simple(df_input, is_train=False):
         df_drop_nodes['high_price_change_percent'] = prev_pct.loc[drop_mask].astype(float).round(4).to_numpy()
         df_drop_nodes['high_price_change_amount'] = prev_amo.loc[drop_mask].astype(float).round(2).to_numpy()
         df_drop_nodes['high_price_amount'] = prev_price.loc[drop_mask].astype(float).round(2).to_numpy()
-        df_drop_nodes['high_price_seats_remaining'] = prev_seats.loc[drop_mask].astype(int).to_numpy()
+        df_drop_nodes['high_price_seats_remaining_transition'] = prev_seats_trans.loc[drop_mask].astype(str)
         df_drop_nodes = df_drop_nodes.reset_index(drop=True)
 
         flight_info_cols = [
@@ -981,11 +1005,15 @@ def preprocess_data_simple(df_input, is_train=False):
         drop_info_cols = ['drop_update_hour', 'drop_days_to_departure',
                           'drop_hours_until_departure', 'drop_price_change_percent', 'drop_price_change_amount',
                           'high_price_duration_hours', 'high_price_change_percent', 'high_price_change_amount',
-                          'high_price_amount', 'high_price_seats_remaining',
+                          'high_price_amount', 'high_price_seats_remaining_transition',
         ]
         # 按顺序排列 保留gid
         df_drop_nodes = df_drop_nodes[flight_info_cols + drop_info_cols]
-        # df_drop_nodes = df_drop_nodes[df_drop_nodes['drop_price_change_percent'] <= -0.01]   # 太低的降幅不计
+        df_drop_nodes['start_hours_until_departure'] = (df_drop_nodes['drop_hours_until_departure'] + df_drop_nodes['high_price_duration_hours']).round().astype('Int64')
+        df_drop_nodes = df_drop_nodes[df_drop_nodes['drop_hours_until_departure'] <= 360]
+        df_drop_nodes = df_drop_nodes[df_drop_nodes['start_hours_until_departure'] >= 72]        
+        df_drop_nodes = df_drop_nodes[df_drop_nodes['high_price_duration_hours'] > 1.0]       # 维持时间太短的不计  
+        df_drop_nodes = df_drop_nodes[df_drop_nodes['drop_price_change_amount'].abs() >= 1]   # 1$之内的降价不计
 
         # 反例库：所有有效节点（不限升价）中，未来24小时内未发生降价
         # seg_start_mask = df_target['price_duration_hours'].eq(1)
@@ -1023,7 +1051,7 @@ def preprocess_data_simple(df_input, is_train=False):
         df_rise_nodes['prev_rise_change_percent'] = prev_pct.loc[rise_mask].astype(float).round(4).to_numpy()
         df_rise_nodes['prev_rise_change_amount'] = prev_amo.loc[rise_mask].astype(float).round(2).to_numpy()
         df_rise_nodes['prev_rise_amount'] = prev_price.loc[rise_mask].astype(float).round(2).to_numpy()
-        df_rise_nodes['prev_rise_seats_remaining'] = prev_seats.loc[rise_mask].astype(int).to_numpy()
+        df_rise_nodes['prev_rise_seats_remaining_transition'] = prev_seats_trans.loc[rise_mask].astype(str)
         df_rise_nodes = df_rise_nodes.reset_index(drop=True)
 
         df_rise_nodes = df_rise_nodes.merge(df_gid_info, on='gid', how='left')
@@ -1031,9 +1059,14 @@ def preprocess_data_simple(df_input, is_train=False):
             'rise_update_hour', 'rise_days_to_departure',
             'rise_hours_until_departure', 'rise_price_change_percent', 'rise_price_change_amount',
             'prev_rise_duration_hours', 'prev_rise_change_percent', 'prev_rise_change_amount',
-            'prev_rise_amount', 'prev_rise_seats_remaining',
+            'prev_rise_amount', 'prev_rise_seats_remaining_transition',
         ]
         df_rise_nodes = df_rise_nodes[flight_info_cols + rise_info_cols]
+        df_rise_nodes['start_hours_until_departure'] = (df_rise_nodes['rise_hours_until_departure'] + df_rise_nodes['prev_rise_duration_hours']).round().astype('Int64')
+        df_rise_nodes = df_rise_nodes[df_rise_nodes['rise_hours_until_departure'] <= 360]
+        df_rise_nodes = df_rise_nodes[df_rise_nodes['start_hours_until_departure'] >= 72]
+        df_rise_nodes = df_rise_nodes[df_rise_nodes['prev_rise_duration_hours'] > 1.0]        # 维持时间太短的不计
+        df_rise_nodes = df_rise_nodes[df_rise_nodes['rise_price_change_amount'].abs() >= 1]   # 1$之内的改变不计
 
         # 制作历史包络线
         envelope_group = ['city_pair', 'flight_number_1', 'flight_number_2', 'flight_day']
@@ -1090,58 +1123,117 @@ def predict_data_simple(df_input, group_route_str, output_dir, predict_dir=".",
         df_rise_nodes = pd.read_csv(rise_info_csv_path)
     else:
         df_rise_nodes = pd.DataFrame()
+    
+    # 联合价格分布 ==========================================================
+    # 统一初始化
+    df_min_hours['relative_position'] = np.nan
+    if not df_drop_nodes.empty:
+        df_drop_nodes['relative_position'] = np.nan
+    if not df_rise_nodes.empty:
+        df_rise_nodes['relative_position'] = np.nan
+
+    parts = []
+
+    # 当前待预测
+    if not df_min_hours.empty and 'adult_total_price' in df_min_hours.columns:
+        cur = df_min_hours[['adult_total_price']].copy()
+        cur['price'] = pd.to_numeric(cur['adult_total_price'], errors='coerce')
+        cur['source'] = 'min'
+        cur['row_id'] = cur.index
+        parts.append(cur[['price', 'source', 'row_id']])
+
+    # 历史降价
+    if not df_drop_nodes.empty and 'high_price_amount' in df_drop_nodes.columns:
+        drop = df_drop_nodes[['high_price_amount']].copy()
+        drop['price'] = pd.to_numeric(drop['high_price_amount'], errors='coerce')
+        drop['source'] = 'drop'
+        drop['row_id'] = drop.index
+        parts.append(drop[['price', 'source', 'row_id']])
+
+    # 历史升价
+    if not df_rise_nodes.empty and 'prev_rise_amount' in df_rise_nodes.columns:
+        rise = df_rise_nodes[['prev_rise_amount']].copy()
+        rise['price'] = pd.to_numeric(rise['prev_rise_amount'], errors='coerce')
+        rise['source'] = 'rise'
+        rise['row_id'] = rise.index
+        parts.append(rise[['price', 'source', 'row_id']])
+    
+    if parts:
+        all_prices = pd.concat(parts, ignore_index=True)
+        all_prices = all_prices.dropna(subset=['price']).reset_index(drop=True)
+
+        # 计算价格百分位
+        dense_rank = all_prices['price'].rank(method='dense')
+        max_rank = dense_rank.max()
+        if pd.notna(max_rank) and max_rank > 1:
+            all_prices['relative_position'] = (dense_rank - 1) / (max_rank - 1)
+        else:
+            all_prices['relative_position'] = 1.0
+        all_prices['relative_position'] = all_prices['relative_position'].round(4)
+
+        # 回填到三个表
+        m = all_prices['source'] == 'min'
+        df_min_hours.loc[all_prices.loc[m, 'row_id'], 'relative_position'] = all_prices.loc[m, 'relative_position'].values
 
-    # ==================== 跨航班日包络线 + 降价潜力 ====================
-    print(">>> 构建跨航班日价格包络线")
-    flight_key = ['city_pair', 'flight_number_1', 'flight_number_2']
-    day_key = flight_key + ['flight_day']
-
-    # 1. 历史侧：加载训练阶段的峰值数据
-    envelope_csv_path = os.path.join(output_dir, f'{group_route_str}_envelope_info.csv')
-    if os.path.exists(envelope_csv_path):
-        df_hist = pd.read_csv(envelope_csv_path)
-        df_hist = df_hist[day_key + ['peak_price', 'peak_hours']]
-        df_hist['source'] = 'hist'
-    else:
-        df_hist = pd.DataFrame()
-
-    # 2. 未来侧：当前在售价格
-    df_future = df_min_hours[day_key + ['adult_total_price', 'hours_until_departure']].copy().rename(
-        columns={'adult_total_price': 'peak_price', 'hours_until_departure': 'peak_hours'}
-    )
-    df_future['source'] = 'future'
-
-    # 3. 合并包络线数据点
-    df_envelope_all = pd.concat(
-        [x for x in [df_hist, df_future] if not x.empty], ignore_index=True
-    ).drop_duplicates(subset=day_key, keep='last')
-
-    # 4. 包络线统计 + 找高点起飞日
-    df_envelope_agg = df_envelope_all.groupby(flight_key).agg(
-        envelope_max=('peak_price', 'max'),               # 峰值最大 
-        envelope_min=('peak_price', 'min'),               # 峰值最小
-        envelope_mean=('peak_price', 'mean'),             # 峰值平均
-        envelope_count=('peak_price', 'count'),           # 峰值统计总数
-        envelope_avg_peak_hours=('peak_hours', 'mean'),   # 峰值发生的距离起飞小时数, 做一下平均
-    ).reset_index()
+        if not df_drop_nodes.empty:
+            m = all_prices['source'] == 'drop'
+            df_drop_nodes.loc[all_prices.loc[m, 'row_id'], 'relative_position'] = all_prices.loc[m, 'relative_position'].values
 
-    # 对数值列保留两位小数
-    df_envelope_agg[['envelope_mean', 'envelope_avg_peak_hours']] = df_envelope_agg[['envelope_mean', 'envelope_avg_peak_hours']].round(2)
+        if not df_rise_nodes.empty:
+            m = all_prices['source'] == 'rise'
+            df_rise_nodes.loc[all_prices.loc[m, 'row_id'], 'relative_position'] = all_prices.loc[m, 'relative_position'].values
 
-    idx_top = df_envelope_all.groupby(flight_key)['peak_price'].idxmax()
-    df_top = df_envelope_all.loc[idx_top, flight_key + ['flight_day', 'peak_price', 'peak_hours']].rename(
-        columns={'flight_day': 'target_flight_day', 'peak_price': 'target_price', 'peak_hours': 'target_peak_hours'}
-    )
-    df_envelope_agg = df_envelope_agg.merge(df_top, on=flight_key, how='left')
+    # ==================== 跨航班日包络线 + 降价潜力 ====================
+    # print(">>> 构建跨航班日价格包络线")
+    # flight_key = ['city_pair', 'flight_number_1', 'flight_number_2']
+    # day_key = flight_key + ['flight_day']
+
+    # # 1. 历史侧：加载训练阶段的峰值数据
+    # envelope_csv_path = os.path.join(output_dir, f'{group_route_str}_envelope_info.csv')
+    # if os.path.exists(envelope_csv_path):
+    #     df_hist = pd.read_csv(envelope_csv_path)
+    #     df_hist = df_hist[day_key + ['peak_price', 'peak_hours']]
+    #     df_hist['source'] = 'hist'
+    # else:
+    #     df_hist = pd.DataFrame()
+
+    # # 2. 未来侧：当前在售价格
+    # df_future = df_min_hours[day_key + ['adult_total_price', 'hours_until_departure']].copy().rename(
+    #     columns={'adult_total_price': 'peak_price', 'hours_until_departure': 'peak_hours'}
+    # )
+    # df_future['source'] = 'future'
+
+    # # 3. 合并包络线数据点
+    # df_envelope_all = pd.concat(
+    #     [x for x in [df_hist, df_future] if not x.empty], ignore_index=True
+    # ).drop_duplicates(subset=day_key, keep='last')
+
+    # # 4. 包络线统计 + 找高点起飞日
+    # df_envelope_agg = df_envelope_all.groupby(flight_key).agg(
+    #     envelope_max=('peak_price', 'max'),               # 峰值最大 
+    #     envelope_min=('peak_price', 'min'),               # 峰值最小
+    #     envelope_mean=('peak_price', 'mean'),             # 峰值平均
+    #     envelope_count=('peak_price', 'count'),           # 峰值统计总数
+    #     envelope_avg_peak_hours=('peak_hours', 'mean'),   # 峰值发生的距离起飞小时数, 做一下平均
+    # ).reset_index()
+
+    # # 对数值列保留两位小数
+    # df_envelope_agg[['envelope_mean', 'envelope_avg_peak_hours']] = df_envelope_agg[['envelope_mean', 'envelope_avg_peak_hours']].round(2)
+
+    # idx_top = df_envelope_all.groupby(flight_key)['peak_price'].idxmax()
+    # df_top = df_envelope_all.loc[idx_top, flight_key + ['flight_day', 'peak_price', 'peak_hours']].rename(
+    #     columns={'flight_day': 'target_flight_day', 'peak_price': 'target_price', 'peak_hours': 'target_peak_hours'}
+    # )
+    # df_envelope_agg = df_envelope_agg.merge(df_top, on=flight_key, how='left')
 
-    # 5. 合并到 df_min_hours
-    df_min_hours = df_min_hours.merge(df_envelope_agg, on=flight_key, how='left')
-    price_range = (df_min_hours['envelope_max'] - df_min_hours['envelope_min']).replace(0, 1)    # 计算当前价格在包络区间的百分位
-    df_min_hours['envelope_position'] = (
-        (df_min_hours['adult_total_price'] - df_min_hours['envelope_min']) / price_range
-    ).clip(0, 1).round(4)
-    # df_min_hours['is_envelope_peak'] = (df_min_hours['envelope_position'] >= 0.75).astype(int)   # 0.95 -> 0.75
-    df_min_hours['is_target_day'] = (df_min_hours['flight_day'] == df_min_hours['target_flight_day']).astype(int)
+    # # 5. 合并到 df_min_hours
+    # df_min_hours = df_min_hours.merge(df_envelope_agg, on=flight_key, how='left')
+    # price_range = (df_min_hours['envelope_max'] - df_min_hours['envelope_min']).replace(0, 1)    # 计算当前价格在包络区间的百分位
+    # df_min_hours['envelope_position'] = (
+    #     (df_min_hours['adult_total_price'] - df_min_hours['envelope_min']) / price_range
+    # ).clip(0, 1).round(4)
+    # # df_min_hours['is_envelope_peak'] = (df_min_hours['envelope_position'] >= 0.75).astype(int)   # 0.95 -> 0.75
+    # df_min_hours['is_target_day'] = (df_min_hours['flight_day'] == df_min_hours['target_flight_day']).astype(int)
 
     # # ==================== 目标二：降价潜力评分 ====================
     # # 用“上涨后回落倾向”替代简单计数：drop / (drop + rise)
@@ -1192,16 +1284,20 @@ def predict_data_simple(df_input, group_route_str, output_dir, predict_dir=".",
     # ).round(4)
 
     # 综合评分阈值：大于阈值的都认为值得投放
-    target_score_threshold = 0.5
+    target_score_threshold = 0.4
     # df_min_hours['target_score_threshold'] = target_score_threshold
-    df_min_hours['is_good_target'] = (df_min_hours['envelope_position'] >= target_score_threshold).astype(int)
+    df_min_hours['is_good_target'] = (df_min_hours['relative_position'] >= target_score_threshold).astype(int)
 
     print(f">>> 包络线+降价潜力评分完成")
-    del df_hist, df_future, df_envelope_all, df_envelope_agg, df_top   # df_drop_freq, df_rise_freq
+    # del df_hist, df_future, df_envelope_all, df_envelope_agg, df_top   # df_drop_freq, df_rise_freq
     
     total_cnt_before = len(df_min_hours)   # 记录下过滤前的总数
-    df_min_hours = df_min_hours[(df_min_hours['is_good_target'] == 1) & (df_min_hours['seats_remaining'] >= 3)].reset_index(drop=True)   # 保留值得投放的 
+    df_min_hours = df_min_hours[(df_min_hours['is_good_target'] == 1) & (df_min_hours['seats_remaining'] >= 1)].reset_index(drop=True)   # 保留值得投放的 
     total_cnt_after = len(df_min_hours)    # 记录下过滤后的总数
+
+    # 余票为1的样本去掉
+    # df_drop_nodes = df_drop_nodes[df_drop_nodes['high_price_seats_remaining'] >= 2]
+    # df_rise_nodes = df_rise_nodes[df_rise_nodes['prev_rise_seats_remaining'] >= 2]
     # =====================================================================
     # df_min_hours = df_min_hours[(df_min_hours['seats_remaining'] >= 5)].reset_index(drop=True)
 
@@ -1240,6 +1336,7 @@ def predict_data_simple(df_input, group_route_str, output_dir, predict_dir=".",
         # seats_remaining_change_amount = row['seats_remaining_change_amount']
         price_amount = row['adult_total_price']
         seats_remaining = row['seats_remaining']
+        seats_remaining_transition = row['seats_remaining_transition']
         # envelope_position = row['envelope_position']
 
         length_drop = 0
@@ -1274,7 +1371,7 @@ def predict_data_simple(df_input, group_route_str, output_dir, predict_dir=".",
                     pct_vals.notna(),
                     ['drop_days_to_departure', 'drop_hours_until_departure', 'drop_price_change_percent', 'drop_price_change_amount',
                      'high_price_duration_hours', 'high_price_change_percent', 
-                     'high_price_change_amount', 'high_price_amount', 'high_price_seats_remaining']
+                     'high_price_change_amount', 'high_price_amount', 'high_price_seats_remaining_transition']
                 ].copy()
                 df_drop_gap['pct_gap'] = (pct_vals.loc[pct_vals.notna()] - pct_base)
                 df_drop_gap['pct_abs_gap'] = df_drop_gap['pct_gap'].abs()
@@ -1291,8 +1388,9 @@ def predict_data_simple(df_input, group_route_str, output_dir, predict_dir=".",
                 )
                 df_match = df_drop_gap[
                     (df_drop_gap['pct_abs_gap'] <= pct_threshold)
-                    & (df_drop_gap['price_abs_gap'] <= 0.1)
+                    & (df_drop_gap['price_abs_gap'] <= 0.001)
                     & same_sign_mask
+                    & (df_drop_gap['high_price_seats_remaining_transition'] == seats_remaining_transition)
                 ].copy()
                 # df_match = df_drop_gap[(df_drop_gap['pct_abs_gap'] <= pct_threshold) & (df_drop_gap['price_abs_gap'] <= 1.0)].copy()
                 # df_drop_gap = df_drop_gap.sort_values(['price_abs_gap'], ascending=[True])
@@ -1392,7 +1490,7 @@ def predict_data_simple(df_input, group_route_str, output_dir, predict_dir=".",
                     pct_vals_1.notna(),
                     ['rise_days_to_departure', 'rise_hours_until_departure', 'rise_price_change_percent', 'rise_price_change_amount',
                      'prev_rise_duration_hours', 'prev_rise_change_percent', 
-                     'prev_rise_change_amount', 'prev_rise_amount', 'prev_rise_seats_remaining']
+                     'prev_rise_change_amount', 'prev_rise_amount', 'prev_rise_seats_remaining_transition']
                 ].copy()
                 df_rise_gap_1['pct_gap'] = (pct_vals_1.loc[pct_vals_1.notna()] - pct_base_1)
                 df_rise_gap_1['pct_abs_gap'] = df_rise_gap_1['pct_gap'].abs()
@@ -1409,8 +1507,9 @@ def predict_data_simple(df_input, group_route_str, output_dir, predict_dir=".",
                 )
                 df_match_1 = df_rise_gap_1[
                     (df_rise_gap_1['pct_abs_gap'] <= pct_threshold_1)
-                    & (df_rise_gap_1['price_abs_gap'] <= 0.1)
+                    & (df_rise_gap_1['price_abs_gap'] <= 0.001)
                     & same_sign_mask_1
+                    & (df_rise_gap_1['prev_rise_seats_remaining_transition'] == seats_remaining_transition)
                 ].copy()
                 # df_match_1 = df_rise_gap_1.loc[(df_rise_gap_1['pct_abs_gap'] <= pct_threshold_1) & (df_rise_gap_1['price_abs_gap'] <= 1.0)].copy()
                 # df_rise_gap_1 = df_rise_gap_1.sort_values(['price_abs_gap'], ascending=[True])
@@ -1532,11 +1631,11 @@ def predict_data_simple(df_input, group_route_str, output_dir, predict_dir=".",
                   'flag_dist',
                   'drop_price_change_upper', 'drop_price_change_lower', 'drop_price_sample_size',
                   'rise_price_change_upper', 'rise_price_change_lower', 'rise_price_sample_size',
-                  'envelope_max', 'envelope_min', 'envelope_mean', 'envelope_count',
-                  'envelope_avg_peak_hours', 'envelope_position',                             # 包络线特征
-                  'target_flight_day', 'target_price', 'target_peak_hours', 'is_target_day',  # 高点起飞日（纯包络线高点）
+                # 'envelope_max', 'envelope_min', 'envelope_mean', 'envelope_count',
+                # 'envelope_avg_peak_hours', 'envelope_position',                             # 包络线特征
+                # 'target_flight_day', 'target_price', 'target_peak_hours', 'is_target_day',  # 高点起飞日（纯包络线高点）
                 # 'drop_freq_count', 'drop_potential', 'target_score',                        # 降价潜力 
-                  'is_good_target',                                                           # 综合目标评分（） 
+                  'relative_position', 'is_good_target',                                      # 综合目标评分（） 
                  ]
     df_predict = df_min_hours[order_cols]
     df_predict = df_predict.rename(columns={

descending_cabin_task.py

@@ -517,7 +517,7 @@ def _process_one_task(row, runner):
         return None
     
     drop_price_sample_size = int(task.get("drop_price_sample_size", "0"))
-    if drop_price_sample_size < 2:  # 丢弃历史降价样本数过少(小于2)的
+    if drop_price_sample_size < 1:  # 丢弃历史降价样本数过少(小于1)的
         return None
     
     flight_day = task.get("flight_day")

main_tr_0.py

@@ -50,7 +50,7 @@ def start_train():
     # date_end = datetime.today().strftime("%Y-%m-%d")
     date_end = (datetime.today() - timedelta(days=1)).strftime("%Y-%m-%d")
     # date_begin = (datetime.today() - timedelta(days=32)).strftime("%Y-%m-%d")
-    date_begin = "2026-03-01"   # 2026-01-01  2026-04-23
+    date_begin = "2026-04-01"   # 2026-03-01  2026-04-28
 
     print(f"训练时间范围: {date_begin} 到 {date_end}")