jiangcang_uo/data_process.py

import pandas as pd
import numpy as np
import gc
import os


def preprocess_data_simple(df_input, is_train=False, hourly_time=None):

    print(">>> 开始数据预处理")
    # 城市码映射成数字(不用)

    # 更新日期是周几
    df_input['update_week'] = df_input['update_hour'].dt.dayofweek + 1
    
    # gid：基于指定字段的分组标记（整数）
    df_input['gid'] = (
        df_input
        .groupby(
            ['citypair', 'flight_numbers', 'from_date'],    # 'baggage_weight' 先不进分组
            sort=False
        )
        .ngroup()
    )

    # 在 gid 与 baggage_weight 内按时间降序
    df_input = df_input.sort_values(
        by=['gid', 'baggage_weight', 'hours_until_departure'],
        ascending=[True, True, False]
    ).reset_index(drop=True)

    df_input = df_input[df_input['hours_until_departure'] <= 480]
    df_input = df_input[df_input['baggage_weight'] == 0]   # 先保留0公斤行李的

    # 在hours_until_departure 的末尾 保留到当前时刻的数据
    if not is_train:
        df_input = df_input[df_input['update_hour'] <= hourly_time].copy()
    else:
        df_input = df_input.copy()  # 训练集也 copy 一下保持一致性
    
    df_input = df_input.reset_index(drop=True)

    # 价格变化最小量阈值
    price_change_amount_threshold = 1
    df_input['_raw_price_diff'] = df_input.groupby(['gid', 'baggage_weight'], group_keys=False)['price_total'].diff()

    # 计算价格变化量
    df_input['price_change_amount'] = (
        df_input['_raw_price_diff']
        .mask(df_input['_raw_price_diff'].abs() < price_change_amount_threshold, 0)
        .replace(0, np.nan)
        .groupby([df_input['gid'], df_input['baggage_weight']], group_keys=False)
        .ffill()
        .fillna(0)
        .round(2)
    )

    # 计算价格变化百分比（相对于上一时间点的变化率）
    df_input['price_change_percent'] = (
        df_input.groupby(['gid', 'baggage_weight'], group_keys=False)['price_total']
        .pct_change()
        .mask(df_input['_raw_price_diff'].abs() < price_change_amount_threshold, 0)
        .replace(0, np.nan)
        .groupby([df_input['gid'], df_input['baggage_weight']], group_keys=False)
        .ffill()
        .fillna(0)
        .round(4)
    )

    # 第一步：标记价格变化段（按“是否发生新的实际变价事件”切段）
    # 这样即使连续两次变价金额相同（如 -50, -50），也会分到不同段
    _price_change_event = df_input['_raw_price_diff'].abs().ge(price_change_amount_threshold)
    df_input['price_change_segment'] = (
        _price_change_event
        .groupby([df_input['gid'], df_input['baggage_weight']], group_keys=False)
        .cumsum()
    )

    # 第二步：计算每个变化段内的持续时间
    df_input['price_duration_hours'] = (
        df_input.groupby(['gid', 'baggage_weight', 'price_change_segment'], group_keys=False)
        .cumcount()
        .add(1)
    )

    # 可选：删除临时列
    df_input = df_input.drop(columns=['price_change_segment', '_raw_price_diff'])

    # 训练过程
    if is_train:
        df_target = df_input[(df_input['hours_until_departure'] >= 48) & (df_input['hours_until_departure'] <= 384)].copy()
        df_target = df_target.sort_values(
            by=['gid', 'baggage_weight', 'hours_until_departure'],
            ascending=[True, True, False]
        ).reset_index(drop=True)

        # 每条对应的前一条记录
        prev_pct = df_target.groupby(['gid', 'baggage_weight'], group_keys=False)['price_change_percent'].shift(1)
        prev_amo = df_target.groupby(['gid', 'baggage_weight'], group_keys=False)['price_change_amount'].shift(1)
        prev_dur = df_target.groupby(['gid', 'baggage_weight'], group_keys=False)['price_duration_hours'].shift(1)
        prev_price = df_target.groupby(['gid', 'baggage_weight'], group_keys=False)['price_total'].shift(1)
        prev_cabin = df_target.groupby(['gid', 'baggage_weight'], group_keys=False)['cabins'].shift(1)

        # 对于先升后降（先降再降）的分析
        seg_start_mask = df_target['price_duration_hours'].eq(1)   # 开始变价节点
        # 正例库仅保留24小时内发生的降价：上一价格段持续时长需<=24h
        prev_pct_num = pd.to_numeric(prev_pct, errors='coerce')
        drop_mask = (
            seg_start_mask
            & prev_pct_num.notna()
            & (df_target['price_change_percent'] < 0)
            & prev_dur.le(24)
        )

        df_drop_nodes = df_target.loc[drop_mask, ['gid', 'baggage_weight', 'hours_until_departure', 'days_to_departure', 'update_hour', 'update_week', 'cabins']].copy()
        df_drop_nodes.rename(columns={'hours_until_departure': 'drop_hours_until_departure'}, inplace=True)
        df_drop_nodes.rename(columns={'days_to_departure': 'drop_days_to_departure'}, inplace=True)
        df_drop_nodes.rename(columns={'update_hour': 'drop_update_hour'}, inplace=True)
        df_drop_nodes.rename(columns={'update_week': 'drop_update_week'}, inplace=True)
        df_drop_nodes.rename(columns={'cabins': 'drop_cabins'}, inplace=True)
        df_drop_nodes['drop_price_change_percent'] = df_target.loc[drop_mask, 'price_change_percent'].astype(float).round(4).to_numpy()
        df_drop_nodes['drop_price_change_amount'] = df_target.loc[drop_mask, 'price_change_amount'].astype(float).round(2).to_numpy()
        df_drop_nodes['high_price_duration_hours'] = prev_dur.loc[drop_mask].astype(float).to_numpy()
        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_cabins'] = prev_cabin.loc[drop_mask].astype(str)
        df_drop_nodes = df_drop_nodes.reset_index(drop=True)

        flight_info_cols = [
            'gid', 'baggage_weight', 'citypair', 'flight_numbers', 'from_time', 'from_date', 'currency',
        ]
        flight_info_cols = [c for c in flight_info_cols if c in df_target.columns]
        df_gid_info = df_target[flight_info_cols].drop_duplicates(subset=['gid', 'baggage_weight']).reset_index(drop=True)
        df_drop_nodes = df_drop_nodes.merge(df_gid_info, on=['gid', 'baggage_weight'], how='left')

        drop_info_cols = [
            'drop_update_hour', 'drop_update_week', 'drop_cabins', 
            '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_cabins',
        ]
        # 按顺序排列 保留gid
        df_drop_nodes = df_drop_nodes[flight_info_cols + drop_info_cols]
        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] 
        
        # 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)
        # rise_mask = seg_start_mask & ((prev_pct > 0) | (prev_pct < 0)) & (df_target['price_change_percent'] > 0)
        prev_pct_num = pd.to_numeric(prev_pct, errors='coerce')
        valid_mask = seg_start_mask & prev_pct_num.notna()

        curr_pct = pd.to_numeric(df_target['price_change_percent'], errors='coerce')
        prev_dur_num = pd.to_numeric(prev_dur, errors='coerce')
        pos_case_mask = curr_pct.ge(0)
        neg_case_mask = curr_pct.lt(0) & prev_dur_num.gt(24)
        rise_mask = valid_mask & (pos_case_mask | neg_case_mask)

        df_rise_nodes = df_target.loc[rise_mask, ['gid', 'baggage_weight', 'hours_until_departure', 'days_to_departure', 'update_hour', 'update_week', 'cabins']].copy()
        df_rise_nodes.rename(columns={'hours_until_departure': 'rise_hours_until_departure'}, inplace=True)
        df_rise_nodes.rename(columns={'days_to_departure': 'rise_days_to_departure'}, inplace=True)
        df_rise_nodes.rename(columns={'update_hour': 'rise_update_hour'}, inplace=True)
        df_rise_nodes.rename(columns={'update_week': 'rise_update_week'}, inplace=True)
        df_rise_nodes.rename(columns={'cabins': 'rise_cabins'}, inplace=True)
        df_rise_nodes['rise_price_change_percent'] = df_target.loc[rise_mask, 'price_change_percent'].astype(float).round(4).to_numpy()
        df_rise_nodes['rise_price_change_amount'] = df_target.loc[rise_mask, 'price_change_amount'].astype(float).round(2).to_numpy()
        df_rise_nodes['prev_rise_duration_hours'] = prev_dur.loc[rise_mask].astype(float).to_numpy()
        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_cabins'] = prev_cabin.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', 'baggage_weight'], how='left')
        
        rise_info_cols = [
            'rise_update_hour', 'rise_update_week', 'rise_cabins', 
            '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_cabins'
        ]
        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]
        
        # 制作历史包络线
        envelope_group = ['citypair', 'flight_numbers', 'from_date', 'baggage_weight']
        idx_peak = df_target.groupby(envelope_group)['price_total'].idxmax()
        df_envelope = df_target.loc[idx_peak, envelope_group + [
            'from_time', 'price_total', 'hours_until_departure', 'days_to_departure', 'update_hour', 'update_week',
        ]].rename(columns={
            'price_total': 'peak_price',
            'hours_until_departure': 'peak_hours',
            'days_to_departure': 'peak_days',
            'update_hour': 'peak_time',
            'update_week': 'peak_week',
        }).reset_index(drop=True)
        
        del df_gid_info
        del df_target

        return df_input, df_drop_nodes, df_rise_nodes, df_envelope
    
    return df_input, None, None, None
 

def predict_data_simple(df_input, city_pair, object_dir, predict_dir=".", pred_time_str=""):
    if df_input is None or df_input.empty:
        return pd.DataFrame()

    df_sorted = df_input.sort_values(
        by=['gid', 'baggage_weight', 'hours_until_departure'],
        ascending=[True, True, False],
    ).reset_index(drop=True)

    df_sorted = df_sorted[
        df_sorted['hours_until_departure'].between(72, 360)
    ].reset_index(drop=True)

    # 每个 gid  baggage_weight 取 hours_until_departure 最小的一条 (当前小时)
    df_min_hours = (
        df_sorted.drop_duplicates(subset=['gid', 'baggage_weight'], keep='last')
        .reset_index(drop=True)
    )

    # 余票不能太少
    df_min_hours = df_min_hours[(df_min_hours['ticket_amount'] >= 2)].reset_index(drop=True)

    # 读历史降价场景
    drop_info_csv_path = os.path.join(object_dir, f'{city_pair}_drop_info.csv')
    if os.path.exists(drop_info_csv_path):
        df_drop_nodes = pd.read_csv(drop_info_csv_path)
    else:
        df_drop_nodes = pd.DataFrame()

    # 读历史升价场景
    rise_info_csv_path = os.path.join(object_dir, f'{city_pair}_rise_info.csv')
    if os.path.exists(rise_info_csv_path):
        df_rise_nodes = pd.read_csv(rise_info_csv_path)
    else:
        df_rise_nodes = pd.DataFrame()
    
    # 联合价格分布（按 flight_key 分组计算）==========================================================
    flight_key = ['citypair', 'flight_numbers', 'baggage_weight']
    group_cols = [
        c for c in flight_key
        if (
            c in df_min_hours.columns
            or (not df_drop_nodes.empty and c in df_drop_nodes.columns)
            or (not df_rise_nodes.empty and c in df_rise_nodes.columns)
        )
    ]

    # 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 'price_total' in df_min_hours.columns:
        min_group_cols = [c for c in group_cols if c in df_min_hours.columns]
        cur = df_min_hours[min_group_cols + ['from_date', 'price_total']].copy()
        for c in group_cols:
            if c not in cur.columns:
                cur[c] = np.nan
        cur['price'] = pd.to_numeric(cur['price_total'], errors='coerce')
        cur['source'] = 'min'
        cur['row_id'] = cur.index
        parts.append(cur[group_cols + ['from_date', 'price', 'source', 'row_id']])
    
    # 历史降价
    if not df_drop_nodes.empty and 'high_price_amount' in df_drop_nodes.columns:
        drop_group_cols = [c for c in group_cols if c in df_drop_nodes.columns]
        drop = df_drop_nodes[drop_group_cols + ['from_date', 'high_price_amount']].copy()
        for c in group_cols:
            if c not in drop.columns:
                drop[c] = np.nan
        drop['price'] = pd.to_numeric(drop['high_price_amount'], errors='coerce')
        drop['source'] = 'drop'
        drop['row_id'] = drop.index
        parts.append(drop[group_cols + ['from_date', 'price', 'source', 'row_id']])
    
    # 历史升价
    if not df_rise_nodes.empty and 'prev_rise_amount' in df_rise_nodes.columns:
        rise_group_cols = [c for c in group_cols if c in df_rise_nodes.columns]
        rise = df_rise_nodes[rise_group_cols + ['from_date', 'prev_rise_amount']].copy()
        for c in group_cols:
            if c not in rise.columns:
                rise[c] = np.nan
        rise['price'] = pd.to_numeric(rise['prev_rise_amount'], errors='coerce')
        rise['source'] = 'rise'
        rise['row_id'] = rise.index
        parts.append(rise[group_cols + ['from_date', '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（优先按分组计算，无法分组时回退全局）
        if group_cols:
            all_prices['dense_rank'] = all_prices.groupby(group_cols, dropna=False)['price'].rank(method='dense')
            all_prices['max_rank'] = all_prices.groupby(group_cols, dropna=False)['dense_rank'].transform('max')
            all_prices['relative_position'] = np.where(
                all_prices['max_rank'] > 1,
                (all_prices['dense_rank'] - 1) / (all_prices['max_rank'] - 1),
                1.0,
            )
            all_prices = all_prices.drop(columns=['dense_rank', 'max_rank'])
        else:    
            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

        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

        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
    
    # ====================================================================================================

    # print(">>> 构建跨航班日价格包络线")
    # flight_key = ['citypair', 'flight_numbers', 'baggage_weight']
    # day_key = flight_key + ['from_date']

    # # 1. 历史侧：加载训练阶段的峰值数据
    # envelope_csv_path = os.path.join(object_dir, f'{city_pair}_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 + ['price_total', 'hours_until_departure']].copy().rename(
    # #     columns={'price_total': 'peak_price', 'hours_until_departure': 'peak_hours'}
    # # )
    # # df_future['source'] = 'future'
    # df_future = pd.DataFrame()

    # # 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 + ['from_date', 'peak_price', 'peak_hours']].rename(
    #     columns={'from_date': '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['price_total'] - 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['from_date'] == df_min_hours['target_flight_day']).astype(int)
    
    # 综合评分阈值：大于阈值的都认为值得投放
    relative_position_threshold = 0.4
    df_min_hours['is_good_target'] = (df_min_hours['relative_position'] >= relative_position_threshold).astype(int)    
    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)].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

    # =====================================================================

    df_min_hours['simple_will_price_drop'] = 0
    # df_min_hours['simple_drop_in_hours'] = 0
    df_min_hours['simple_drop_in_hours_prob'] = 0.0
    df_min_hours['simple_drop_in_hours_dist'] = ''   # 空串 表示未知
    df_min_hours['flag_dist'] = ''
    df_min_hours['drop_price_change_upper'] = 0.0
    df_min_hours['drop_price_change_lower'] = 0.0
    df_min_hours['drop_price_sample_size'] = 0
    df_min_hours['rise_price_change_upper'] = 0.0
    df_min_hours['rise_price_change_lower'] = 0.0
    df_min_hours['rise_price_sample_size'] = 0

    # 这个阈值取多少?
    pct_threshold = 0.1
    pct_threshold_1 = 0.1

    for idx, row in df_min_hours.iterrows(): 
        city_pair = row['citypair']
        flight_numbers = row['flight_numbers']
        baggage_weight = row['baggage_weight']
        from_date = row['from_date']
        if flight_numbers == "UO235" and from_date == "2026-04-25":   # 调试时用
            pass
        days_to_departure = row['days_to_departure']
        hours_until_departure = row['hours_until_departure']
        price_change_percent = row['price_change_percent']
        price_change_amount = row['price_change_amount']
        price_duration_hours = row['price_duration_hours']
        price_amount = row['price_total']
        cabins = row['cabins']

        length_drop = 0
        length_rise = 0

        # 针对历史上发生的 >降价
        if not df_drop_nodes.empty:
            # 对准航线 航班号 行李配额
            df_drop_nodes_part = df_drop_nodes[
                (df_drop_nodes['citypair'] == city_pair) &
                (df_drop_nodes['flight_numbers'] == flight_numbers) &
                (df_drop_nodes['baggage_weight'] == baggage_weight)
            ]
            # 降价前 增量阈值、当前阈值 的匹配
            if not df_drop_nodes_part.empty and pd.notna(price_change_percent):   
                
                pct_base = float(price_change_percent)
                pct_vals = pd.to_numeric(df_drop_nodes_part['high_price_change_percent'], errors='coerce')
                df_drop_gap = df_drop_nodes_part.loc[
                    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_cabins', 'relative_position'
                     ]
                ].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()

                price_base = pd.to_numeric(price_amount, errors='coerce')
                high_price_vals = pd.to_numeric(df_drop_gap['high_price_amount'], errors='coerce')
                df_drop_gap['price_gap'] = high_price_vals - price_base
                df_drop_gap['price_abs_gap'] = df_drop_gap['price_gap'].abs()

                df_drop_gap = df_drop_gap.sort_values(['price_abs_gap', 'pct_abs_gap'], ascending=[True, True])
                same_sign_mask = (
                    np.sign(pd.to_numeric(df_drop_gap['high_price_change_percent'], errors='coerce'))
                    == np.sign(pct_base)
                )
                df_match = df_drop_gap[
                    (df_drop_gap['pct_abs_gap'] <= pct_threshold) 
                    & (df_drop_gap['price_abs_gap'] <= 5.0)
                    & same_sign_mask
                    & (df_drop_gap['high_price_cabins'] == cabins)
                ].copy()

                # 历史上出现的极近似的增长(下降)幅度后的降价场景
                if not df_match.empty:
                    dur_base = pd.to_numeric(price_duration_hours, errors='coerce')
                    hud_base = pd.to_numeric(hours_until_departure, errors='coerce')
                    dtd_base = pd.to_numeric(days_to_departure, errors='coerce')

                    if pd.notna(dur_base) and pd.notna(dtd_base) and pd.notna(hud_base): 
                        df_match_chk = df_match.copy()

                        # drop_dtd_vals = pd.to_numeric(df_match_chk['drop_days_to_departure'], errors='coerce')
                        # df_match_chk = df_match_chk.loc[drop_dtd_vals.notna()].copy()
                        # df_match_chk = df_match_chk.loc[(drop_dtd_vals.loc[drop_dtd_vals.notna()] - float(dtd_base)).abs() <= 3].copy()

                        # drop_hud_vals = pd.to_numeric(df_match_chk['drop_hours_until_departure'], errors='coerce')
                        # df_match_chk = df_match_chk.loc[drop_hud_vals.notna()].copy()
                        # df_match_chk = df_match_chk.loc[(float(hud_base) - drop_hud_vals.loc[drop_hud_vals.notna()]) >= -24].copy()

                        # 正例收紧
                        dur_num_chk = pd.to_numeric(df_match_chk['high_price_duration_hours'], errors='coerce')
                        dur_delta = dur_num_chk - float(dur_base)
                        df_match_chk = df_match_chk.assign(dur_delta=dur_delta)
                        df_match_chk = df_match_chk.loc[df_match_chk['dur_delta'].notna()].copy()
                        df_match_chk = df_match_chk.loc[df_match_chk['dur_delta'].abs() <= 72].copy()

                        # 所有条件都对的上
                        if not df_match_chk.empty:
                            length_drop = df_match_chk.shape[0]
                            df_min_hours.loc[idx, 'drop_price_sample_size'] = length_drop

                            drop_price_change_upper = df_match_chk['drop_price_change_amount'].max()   # 降价上限
                            drop_price_change_lower = df_match_chk['drop_price_change_amount'].min()   # 降价下限
                            df_min_hours.loc[idx, 'drop_price_change_upper'] = round(drop_price_change_upper, 2)
                            df_min_hours.loc[idx, 'drop_price_change_lower'] = round(drop_price_change_lower, 2)

                            # remaining_hours = (
                            #     pd.to_numeric(df_match_chk['high_price_duration_hours'], errors='coerce') - float(dur_base)
                            # ).clip(lower=0)
                            # remaining_hours = remaining_hours.round().astype(int)

                            # counts = remaining_hours.value_counts().sort_index()
                            # probs = (counts / counts.sum()).round(4)

                            # top_hours = int(probs.idxmax())
                            # top_prob = float(probs.max())

                            # dist_items = list(zip(probs.index.tolist(), probs.tolist()))
                            # dist_items = dist_items[:10]
                            # dist_str = ' '.join([f"{int(h)}h->{float(p)}" for h, p in dist_items])

                            dur_delta_list = df_match_chk['dur_delta'].tolist()
                            dist_str = "'" + ' '.join([f"{ddl:g}" for ddl in dur_delta_list])

                            df_min_hours.loc[idx, 'simple_will_price_drop'] = 1
                            # df_min_hours.loc[idx, 'simple_drop_in_hours'] = top_hours
                            df_min_hours.loc[idx, 'simple_drop_in_hours_prob'] = 1
                            df_min_hours.loc[idx, 'simple_drop_in_hours_dist'] = dist_str
                            df_min_hours.loc[idx, 'flag_dist'] = 'd0'
                    pass
                pass
        
        # 针对历史上发生的 <升价
        if not df_rise_nodes.empty:
            # 对准航线 航班号 行李配额
            df_rise_nodes_part = df_rise_nodes[
                (df_rise_nodes['citypair'] == city_pair) &
                (df_rise_nodes['flight_numbers'] == flight_numbers) &
                (df_rise_nodes['baggage_weight'] == baggage_weight)
            ]
            # 升价前 增量阈值、当前阈值 的匹配
            if not df_rise_nodes_part.empty and pd.notna(price_change_percent):
                pct_base_1 = float(price_change_percent)
                pct_vals_1 = pd.to_numeric(df_rise_nodes_part['prev_rise_change_percent'], errors='coerce')
                df_rise_gap_1 = df_rise_nodes_part.loc[
                    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_cabins', 'relative_position']
                ].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()

                price_base_1 = pd.to_numeric(price_amount, errors='coerce')
                rise_price_vals_1 = pd.to_numeric(df_rise_gap_1['prev_rise_amount'], errors='coerce')
                df_rise_gap_1['price_gap'] = rise_price_vals_1 - price_base_1
                df_rise_gap_1['price_abs_gap'] = df_rise_gap_1['price_gap'].abs()

                df_rise_gap_1 = df_rise_gap_1.sort_values(['price_abs_gap', 'pct_abs_gap'], ascending=[True, True])
                same_sign_mask_1 = (
                    np.sign(pd.to_numeric(df_rise_gap_1['prev_rise_change_percent'], errors='coerce'))
                    == np.sign(pct_base_1)
                )
                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'] <= 5.0)
                    & same_sign_mask_1
                    & (df_rise_gap_1['prev_rise_cabins'] == cabins)
                ].copy()

                # 历史上出现的极近似的增长(下降)幅度后的升价场景
                if not df_match_1.empty:
                    dur_base_1 = pd.to_numeric(price_duration_hours, errors='coerce')
                    hud_base_1 = pd.to_numeric(hours_until_departure, errors='coerce')
                    dtd_base_1 = pd.to_numeric(days_to_departure, errors='coerce')

                    if pd.notna(dur_base_1) and pd.notna(dtd_base_1) and pd.notna(hud_base_1): 
                        df_match_chk_1 = df_match_1.copy()
                        
                        # drop_dtd_vals_1 = pd.to_numeric(df_match_chk_1['rise_days_to_departure'], errors='coerce')
                        # df_match_chk_1 = df_match_chk_1.loc[drop_dtd_vals_1.notna()].copy()
                        # df_match_chk_1 = df_match_chk_1.loc[(drop_dtd_vals_1.loc[drop_dtd_vals_1.notna()] - float(dtd_base_1)).abs() <= 3].copy()

                        # rise_hud_vals_1 = pd.to_numeric(df_match_chk_1['rise_hours_until_departure'], errors='coerce')
                        # df_match_chk_1 = df_match_chk_1.loc[rise_hud_vals_1.notna()].copy()
                        # df_match_chk_1 = df_match_chk_1.loc[(float(hud_base_1) - rise_hud_vals_1.loc[rise_hud_vals_1.notna()]) >= -24].copy()

                        # 反例收紧：48小时内发生降价的不算显著反例
                        _rise_pct_chk = pd.to_numeric(df_match_chk_1['rise_price_change_percent'], errors='coerce')
                        _prev_dur_chk = pd.to_numeric(df_match_chk_1['prev_rise_duration_hours'], errors='coerce')
                        _exclude_mask = _rise_pct_chk.lt(0) & _prev_dur_chk.lt(48)
                        df_match_chk_1 = df_match_chk_1.loc[~_exclude_mask.fillna(False)].copy()

                        # 所有条件都对的上
                        if not df_match_chk_1.empty:
                            length_rise = df_match_chk_1.shape[0]
                            df_min_hours.loc[idx, 'rise_price_sample_size'] = length_rise

                            rise_price_change_upper = df_match_chk_1['rise_price_change_amount'].max()   # 涨价上限
                            rise_price_change_lower = df_match_chk_1['rise_price_change_amount'].min()   # 涨价下限
                            df_min_hours.loc[idx, 'rise_price_change_upper'] = round(rise_price_change_upper, 2)
                            df_min_hours.loc[idx, 'rise_price_change_lower'] = round(rise_price_change_lower, 2)

                            # 可以明确的判定不降价
                            if length_drop == 0:
                                df_min_hours.loc[idx, 'simple_will_price_drop'] = 0
                                # df_min_hours.loc[idx, 'simple_drop_in_hours'] = 0
                                df_min_hours.loc[idx, 'simple_drop_in_hours_prob'] = 0.0
                                # df_min_hours.loc[idx, 'simple_drop_in_hours_dist'] = 'r0'
                                df_min_hours.loc[idx, 'flag_dist'] = 'r0'
                            # 分歧判定
                            else:
                                drop_prob = round(length_drop / (length_rise + length_drop), 2)
                                # 依旧保持之前的降价判定，概率修改
                                if drop_prob > 0.5:
                                    df_min_hours.loc[idx, 'simple_will_price_drop'] = 1
                                    # df_min_hours.loc[idx, 'simple_drop_in_hours_dist'] = 'd1'
                                    df_min_hours.loc[idx, 'flag_dist'] = 'd1'
                                # 改判不降价，概率修改
                                else:
                                    df_min_hours.loc[idx, 'simple_will_price_drop'] = 0
                                    # df_min_hours.loc[idx, 'simple_drop_in_hours_dist'] = 'r1'
                                    df_min_hours.loc[idx, 'flag_dist'] = 'r1'

                                df_min_hours.loc[idx, 'simple_drop_in_hours_prob'] = drop_prob

    print("判定循环结束")

    _dep_hour = pd.to_datetime(df_min_hours["from_time"], errors="coerce").dt.floor("h")
    df_min_hours["valid_begin_hour"] = (_dep_hour - pd.to_timedelta(360, unit="h")).dt.strftime("%Y-%m-%d %H:%M:%S")
    df_min_hours["valid_end_hour"] = (_dep_hour - pd.to_timedelta(72, unit="h")).dt.strftime("%Y-%m-%d %H:%M:%S")

    # 要展示在预测表里的字段
    order_cols = [
        "citypair", "flight_numbers", "baggage_weight", "from_date", "from_time",
        "cabins", "ticket_amount", "currency", "price_base", "price_tax",
        "price_total", 'relative_position', 'is_good_target', 'days_to_departure', 'hours_until_departure', 
        'price_change_amount', 'price_change_percent', 'price_duration_hours', 
        "update_hour", "create_time",
        'valid_begin_hour', 'valid_end_hour',
        'simple_will_price_drop', 'simple_drop_in_hours_prob', 'simple_drop_in_hours_dist',
        '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',
    ]
    df_predict = df_min_hours[order_cols]
    df_predict = df_predict.rename(columns={
            'simple_will_price_drop': 'will_price_drop',
            'simple_drop_in_hours_prob': 'drop_in_hours_prob',
            'simple_drop_in_hours_dist': 'drop_in_hours_dist',
        }
    )

    # 排序
    df_predict = df_predict.sort_values(
        by=['citypair', 'flight_numbers', 'baggage_weight', 'from_date'],
        kind='mergesort',
        na_position='last',
    ).reset_index(drop=True)

    total_cnt = len(df_predict)
    if "will_price_drop" in df_predict.columns:
        _wpd = pd.to_numeric(df_predict["will_price_drop"], errors="coerce")
        drop_1_cnt = int((_wpd == 1).sum())
        drop_0_cnt = int((_wpd == 0).sum())
    else:
        drop_1_cnt = 0
        drop_0_cnt = 0
    print(f"will_price_drop 分类数量统计: 1(会降)={drop_1_cnt}, 0(不降)={drop_0_cnt}, 总数={total_cnt}, 过滤前总数={total_cnt_before}")

    csv_path1 = os.path.join(predict_dir, f'future_predictions_{pred_time_str}.csv')
    df_predict.to_csv(csv_path1, mode='a', index=False, header=not os.path.exists(csv_path1), encoding='utf-8-sig')

    print("预测结果已追加")
    return df_predict