기존 방문객 조회, 전년도 데이터 비교, 예상 방문객 계산

lib/weekly_visitor_forecast.py

@@ -0,0 +1,242 @@
+# weekly_visitor_forecast.py
+import os, sys
+sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
+
+from datetime import date, timedelta, datetime
+from collections import defaultdict
+import pandas as pd
+from sqlalchemy import select, func
+
+from weather_forecast import get_weekly_precip
+from conf import db, db_schema
+from lib.holiday import is_korean_holiday
+from lib.common import load_config
+
+config = load_config()
+visitor_ca_filter = config.get('POS', {}).get('VISITOR_CA', [])
+
+ga4_by_date = db_schema.ga4_by_date
+weather = db_schema.weather
+air = db_schema.air
+pos = db_schema.pos
+
+engine = db.engine 
+
+
+def get_recent_dates(today=None, days=14):
+    today = today or date.today()
+    return [today - timedelta(days=i) for i in reversed(range(days))]
+
+
+def get_this_week_dates(today=None):
+    today = today or date.today()
+    weekday = today.weekday()
+    return [today + timedelta(days=i) for i in range(7 - weekday)]
+
+
+def get_last_year_same_weekdays(dates):
+    return [d.replace(year=d.year - 1) for d in dates]
+
+
+def pm25_grade(value):
+    if value is None:
+        return ''
+    if value <= 15:
+        return '좋음'
+    elif value <= 35:
+        return '보통'
+    elif value <= 75:
+        return '나쁨'
+    else:
+        return '매우나쁨'
+
+
+def fetch_data_for_dates(date_list):
+    session = db.get_session()
+    data = defaultdict(dict)
+    try:
+        # GA4 activeUsers
+        stmt = (
+            select(ga4_by_date.c.date, func.sum(ga4_by_date.c.activeUsers))
+            .where(ga4_by_date.c.date.in_(date_list))
+            .group_by(ga4_by_date.c.date)
+        )
+        for d, val in session.execute(stmt):
+            data[d]['웹 방문자 수'] = val
+
+        # POS 입장객 수
+        stmt = (
+            select(pos.c.date, func.sum(pos.c.qty))
+            .where(
+                (pos.c.date.in_(date_list)) &
+                (pos.c.ca01 == '매표소') &
+                (pos.c.ca03.in_(visitor_ca_filter))
+            )
+            .group_by(pos.c.date)
+        )
+        for d, val in session.execute(stmt):
+            data[d]['입장객 수'] = val
+
+        # 날씨 정보
+        stmt = (
+            select(
+                weather.c.date,
+                func.min(weather.c.minTa),
+                func.max(weather.c.maxTa),
+                func.avg(weather.c.avgRhm),
+                func.sum(weather.c.sumRn)
+            )
+            .where(weather.c.date.in_(date_list))
+            .group_by(weather.c.date)
+        )
+        for row in session.execute(stmt):
+            d, minTa, maxTa, rhm, rn = row
+            data[d]['최저기온'] = round(minTa or 0, 1)
+            data[d]['최고기온'] = round(maxTa or 0, 1)
+            data[d]['습도'] = round(rhm or 0, 1)
+            data[d]['강수량'] = round(rn or 0, 1)
+
+        # 미세먼지 (pm25)
+        stmt = (
+            select(air.c.date, func.avg(air.c.pm25))
+            .where(air.c.date.in_(date_list))
+            .group_by(air.c.date)
+        )
+        for d, pm25 in session.execute(stmt):
+            data[d]['미세먼지'] = pm25_grade(pm25)
+
+    finally:
+        session.close()
+
+    return data
+
+
+def load_prophet_forecast(file_path=None):
+    if file_path is None:
+        file_path = os.path.abspath(os.path.join(os.path.dirname(__file__), '..', 'data', 'prophet_result.csv'))
+    print(f"[DEBUG] Load prophet forecast from: {file_path}")
+
+    if not os.path.exists(file_path):
+        print(f"[ERROR] 파일이 존재하지 않습니다: {file_path}")
+        return pd.Series(dtype=float)
+
+    try:
+        df = pd.read_csv(file_path)
+        # 컬럼명 출력 확인
+        print(f"[DEBUG] CSV columns: {df.columns.tolist()}")
+
+        if 'date' not in df.columns or 'visitor_forecast' not in df.columns:
+            print("[ERROR] 필요한 컬럼이 CSV에 없습니다.")
+            return pd.Series(dtype=float)
+
+        df['date'] = pd.to_datetime(df['date'])
+        df.set_index('date', inplace=True)
+
+        return df['visitor_forecast']
+    except Exception as e:
+        print(f"[ERROR] Prophet 예측 결과 불러오기 실패: {e}")
+        return pd.Series(dtype=float)
+
+
+def build_dataframe(dates, data, use_forecast_after=None):
+    """
+    use_forecast_after: datetime.date or None
+        지정한 날짜 이후부터는 '예상 방문자'로 '입장객 수' 대체
+    """
+    records = []
+    for d in dates:
+        predicted = data.get(d, {}).get('예상 방문자')
+        if use_forecast_after is not None and d >= use_forecast_after and predicted is not None:
+            입장객수 = int(predicted)
+        else:
+            입장객수 = data.get(d, {}).get('입장객 수', 0)
+
+        r = {
+            '날짜': d.strftime('%Y-%m-%d'),
+            '요일': ['월', '화', '수', '목', '금', '토', '일'][d.weekday()],
+            '공휴일': '✅' if is_korean_holiday(d) else '',
+            '웹 방문자 수': data.get(d, {}).get('웹 방문자 수', 0),
+            '입장객 수': 입장객수,
+            '최저기온': data.get(d, {}).get('최저기온', ''),
+            '최고기온': data.get(d, {}).get('최고기온', ''),
+            '습도': data.get(d, {}).get('습도', ''),
+            '강수량': data.get(d, {}).get('강수량', ''),
+            '미세먼지': data.get(d, {}).get('미세먼지', ''),
+        }
+        records.append(r)
+    return pd.DataFrame(records)
+
+
+def main():
+    today = date.today()
+
+    # 이번 주 일요일 (주말)
+    weekday = today.weekday()
+    sunday = today + timedelta(days=(6 - weekday))
+
+    # 최근 2주 및 작년 동일 요일 (최근 2주는 sunday까지 포함)
+    recent_dates = [sunday - timedelta(days=i) for i in reversed(range(14))]
+    prev_year_dates = get_last_year_same_weekdays(recent_dates)
+
+    # 이번 주 예상 대상 (오늘부터 일요일까지)
+    this_week_dates = [today + timedelta(days=i) for i in range(7 - weekday)]
+
+    # 데이터 조회
+    recent_data = fetch_data_for_dates(recent_dates)
+    prev_year_data = fetch_data_for_dates(prev_year_dates)
+    forecast_data = fetch_data_for_dates(this_week_dates)
+
+    # 결측 강수량 보정 - 오늘 이후 날짜가 비어있거나 강수량 없으면 날씨예보로 채움
+    weekly_precip = get_weekly_precip(load_config()['DATA_API']['serviceKey'])
+    for d in recent_dates:
+        if d >= today and (d not in recent_data or '강수량' not in recent_data[d]):
+            dt_str = d.strftime('%Y%m%d')
+            if dt_str in weekly_precip:
+                if d not in recent_data:
+                    recent_data[d] = {}
+                recent_data[d]['강수량'] = round(float(weekly_precip[dt_str]['sumRn']), 1)
+                recent_data[d]['최저기온'] = round(float(weekly_precip[dt_str]['minTa']), 1)
+                recent_data[d]['최고기온'] = round(float(weekly_precip[dt_str]['maxTa']), 1)
+                recent_data[d]['습도'] = round(float(weekly_precip[dt_str]['avgRhm']), 1)
+
+    # prophet 예측 결과 불러오기 및 이번 주 예상 데이터에 병합
+    prophet_forecast = load_prophet_forecast()
+    for d in this_week_dates:
+        d_ts = pd.Timestamp(d)
+        has_forecast = d_ts in prophet_forecast.index
+        print(f"[DEBUG] 날짜 {d} (Timestamp {d_ts}) 예측 데이터 존재 여부: {has_forecast}")
+        if has_forecast:
+            if d not in forecast_data:
+                forecast_data[d] = {}
+            forecast_data[d]['예상 방문자'] = round(float(prophet_forecast.loc[d_ts]), 0)
+        else:
+            if d not in forecast_data:
+                forecast_data[d] = {}
+            forecast_data[d]['예상 방문자'] = None
+
+    # 최근 2주 데이터에도 오늘 이후 날짜에 대해 예상 방문자 병합
+    for d in recent_dates:
+        d_ts = pd.Timestamp(d)
+        if d >= today and d_ts in prophet_forecast.index:
+            if d not in recent_data:
+                recent_data[d] = {}
+            recent_data[d]['예상 방문자'] = round(float(prophet_forecast.loc[d_ts]), 0)
+
+    # 데이터프레임 생성
+    df_recent = build_dataframe(recent_dates, recent_data, use_forecast_after=today)
+    df_prev = build_dataframe(prev_year_dates, prev_year_data)
+
+    # 출력 설정
+    pd.set_option('display.unicode.east_asian_width', True)
+    pd.set_option('display.max_columns', None)
+    pd.set_option('display.width', 200)
+
+    print("📊 최근 2주간 방문자 현황:")
+    print(df_recent.to_string(index=False))
+
+    print("\n📈 작년 동일 요일 데이터:")
+    print(df_prev.to_string(index=False))
+
+
+if __name__ == "__main__":
+    main()