#!/usr/bin/env python3
"""拟合版新增注册用户数：LOESS + 星期因子修正"""
import numpy as np
from datetime import date, timedelta
from collections import defaultdict

# ---- 原始每日新增数据 ----
raw_data = """2025-09-01,3
2025-09-02,10
2025-09-03,4
2025-09-04,5
2025-09-05,11
2025-09-06,8
2025-09-07,16
2025-09-08,11
2025-09-09,137
2025-09-10,63
2025-09-11,26
2025-09-12,27
2025-09-13,41
2025-09-14,39
2025-09-15,27
2025-09-16,57
2025-09-17,58
2025-09-18,55
2025-09-19,133
2025-09-20,104
2025-09-21,101
2025-09-22,132
2025-09-23,126
2025-09-24,71
2025-09-25,37
2025-09-26,34
2025-09-27,81
2025-09-28,35
2025-09-29,47
2025-09-30,30
2025-10-01,48
2025-10-02,62
2025-10-03,45
2025-10-04,42
2025-10-05,45
2025-10-06,62
2025-10-07,42
2025-10-08,45
2025-10-09,36
2025-10-10,62
2025-10-11,90
2025-10-12,93
2025-10-13,162
2025-10-14,131
2025-10-15,112
2025-10-16,131
2025-10-17,215
2025-10-18,129
2025-10-19,81
2025-10-20,44
2025-10-21,41
2025-10-22,45
2025-10-23,37
2025-10-24,56
2025-10-25,79
2025-10-26,50
2025-10-27,89
2025-10-28,86
2025-10-29,82
2025-10-30,92
2025-10-31,78
2025-11-01,129
2025-11-02,168
2025-11-03,77
2025-11-04,68
2025-11-05,48
2025-11-06,67
2025-11-07,177
2025-11-08,160
2025-11-09,105
2025-11-10,78
2025-11-11,72
2025-11-12,234
2025-11-13,104
2025-11-14,69
2025-11-15,89
2025-11-16,63
2025-11-17,70
2025-11-18,82
2025-11-19,155
2025-11-20,71
2025-11-21,90
2025-11-22,71
2025-11-23,91
2025-11-24,47
2025-11-25,77
2025-11-26,93
2025-11-27,92
2025-11-28,77
2025-11-29,122
2025-11-30,125
2025-12-01,91
2025-12-02,94
2025-12-03,188
2025-12-04,131
2025-12-05,125
2025-12-06,190
2025-12-07,194
2025-12-08,108
2025-12-09,116
2025-12-10,110
2025-12-11,104
2025-12-12,120
2025-12-13,190
2025-12-14,172
2025-12-15,98
2025-12-16,100
2025-12-17,97
2025-12-18,85
2025-12-19,142
2025-12-20,127
2025-12-21,131
2025-12-22,69
2025-12-23,77
2025-12-24,108
2025-12-25,84
2025-12-26,89
2025-12-27,107
2025-12-28,95
2025-12-29,50
2025-12-30,67
2025-12-31,64
2026-01-01,77
2026-01-02,74
2026-01-03,69
2026-01-04,42
2026-01-05,56
2026-01-06,33
2026-01-07,52
2026-01-08,59
2026-01-09,58
2026-01-10,83
2026-01-11,75
2026-01-12,34
2026-01-13,46
2026-01-14,59
2026-01-15,31
2026-01-16,31
2026-01-17,66
2026-01-18,71
2026-01-19,53
2026-01-20,48
2026-01-21,40
2026-01-22,62
2026-01-23,46
2026-01-24,72
2026-01-25,86
2026-01-26,61
2026-01-27,57
2026-01-28,146
2026-01-29,102
2026-01-30,87
2026-01-31,61
2026-02-01,54
2026-02-02,53
2026-02-03,42
2026-02-04,39
2026-02-05,42
2026-02-06,55
2026-02-07,36
2026-02-08,47
2026-02-09,40
2026-02-10,60
2026-02-11,265
2026-02-12,59
2026-02-13,42
2026-02-14,31
2026-02-15,41
2026-02-16,28
2026-02-17,52
2026-02-18,23
2026-02-19,34
2026-02-20,26
2026-02-21,36
2026-02-22,35
2026-02-23,46
2026-02-24,45
2026-02-25,64
2026-02-26,180
2026-02-27,101
2026-02-28,167
2026-03-01,124
2026-03-02,106
2026-03-03,76
2026-03-04,105
2026-03-05,447
2026-03-06,216
2026-03-07,239
2026-03-08,206
2026-03-09,167
2026-03-10,108
2026-03-11,131
2026-03-12,177
2026-03-13,243
2026-03-14,131
2026-03-15,131
2026-03-16,108
2026-03-17,94
2026-03-18,95
2026-03-19,77
2026-03-20,103
2026-03-21,133
2026-03-22,126
2026-03-23,84
2026-03-24,80
2026-03-25,94
2026-03-26,85
2026-03-27,95
2026-03-28,104
2026-03-29,101
2026-03-30,81
2026-03-31,99
2026-04-01,138
2026-04-02,162
2026-04-03,771
2026-04-04,340
2026-04-05,185
2026-04-06,206
2026-04-07,212
2026-04-08,749
2026-04-09,336
2026-04-10,128
2026-04-11,174
2026-04-12,150
2026-04-13,116
2026-04-14,132
2026-04-15,126
2026-04-16,115
2026-04-17,84
2026-04-18,117
2026-04-19,119
2026-04-20,88
2026-04-21,97
2026-04-22,179
2026-04-23,139
2026-04-24,121
2026-04-25,140
2026-04-26,137
2026-04-27,120
2026-04-28,163
2026-04-29,65
2026-04-30,58
2026-05-01,63
2026-05-02,60
2026-05-03,60
2026-05-04,52
2026-05-05,70
2026-05-06,98
2026-05-07,133
2026-05-08,86
2026-05-09,89
2026-05-10,106
2026-05-11,98
2026-05-12,82
2026-05-13,99
2026-05-14,98
2026-05-15,166
2026-05-16,116
2026-05-17,123
2026-05-18,88
2026-05-19,115
2026-05-20,132
2026-05-21,130
2026-05-22,127
2026-05-23,200
2026-05-24,121
2026-05-25,67
2026-05-26,104
2026-05-27,99
2026-05-28,112
2026-05-29,107
2026-05-30,121
2026-05-31,178"""

# 解析数据
data = {}
for line in raw_data.strip().split('\n'):
    dt, nu = line.split(',')
    data[dt] = int(nu)

dates = sorted(data.keys())
start_date = date.fromisoformat(dates[0])
end_date = date.fromisoformat(dates[-1])

# ---- 活动日历（活动日+余波日） ----
activity_dates = set()
# 2025: 9/9-10, 9/19-23, 10/13-14, 10/16-17, 11/2, 11/7, 11/10, 11/12, 11/19, 12/3
for d in ['2025-09-09','2025-09-10','2025-09-19','2025-09-20','2025-09-21','2025-09-22','2025-09-23',
          '2025-10-13','2025-10-14','2025-10-16','2025-10-17',
          '2025-11-02','2025-11-07','2025-11-10','2025-11-12','2025-11-19',
          '2025-12-03']:
    activity_dates.add(d)
# 2026: 1/28-29(余波1天), 2/11, 2/26-3/2(余波4天), 3/5-8(余波3天), 3/9, 3/12-13, 4/3-7(余波4天), 4/8-10(余波2天), 4/22-23(余波1天), 4/28, 5/6-7
for d in ['2026-01-28','2026-01-29',
          '2026-02-11',
          '2026-02-26','2026-02-27','2026-02-28','2026-03-01','2026-03-02',
          '2026-03-05','2026-03-06','2026-03-07','2026-03-08',
          '2026-03-09',
          '2026-03-12','2026-03-13',
          '2026-04-03','2026-04-04','2026-04-05','2026-04-06','2026-04-07',
          '2026-04-08','2026-04-09','2026-04-10',
          '2026-04-22','2026-04-23',
          '2026-04-28',
          '2026-05-06','2026-05-07']:
    activity_dates.add(d)

print(f"活动日+余波日总数: {len(activity_dates)}")

# ---- LOESS 实现 ----
def loess(x, y, x_eval, frac=0.236):
    """简单 LOESS 平滑"""
    n = len(x)
    k = max(int(n * frac), 3)
    y_smooth = np.zeros(len(x_eval))
    
    for i, xi in enumerate(x_eval):
        # 计算距离
        dists = np.abs(x - xi)
        idx = np.argsort(dists)[:k]
        max_dist = dists[idx[-1]]
        
        # tricube 权重
        weights = np.zeros(n)
        for j in idx:
            u = dists[j] / max_dist if max_dist > 0 else 0
            weights[j] = (1 - u**3)**3 if u < 1 else 0
        
        # 加权线性回归
        w_sum = weights.sum()
        if w_sum > 0:
            x_w = (x * weights).sum() / w_sum
            y_w = (y * weights).sum() / w_sum
            xy_w = (x * y * weights).sum() / w_sum
            x2_w = (x * x * weights).sum() / w_sum
            
            denom = x2_w - x_w**2
            if abs(denom) > 1e-10:
                b = (xy_w - x_w * y_w) / denom
                a = y_w - b * x_w
                y_smooth[i] = a + b * xi
            else:
                y_smooth[i] = y_w
        else:
            y_smooth[i] = y.mean()
    
    return y_smooth

# 准备数据
x_all = np.arange(len(dates))
y_all = np.array([data[d] for d in dates])

# 清洁日（非活动日）的索引
clean_idx = [i for i, d in enumerate(dates) if d not in activity_dates]
x_clean = x_all[clean_idx]
y_clean = y_all[clean_idx]

print(f"清洁日数量: {len(clean_idx)} / {len(dates)} ({len(clean_idx)/len(dates)*100:.1f}%)")

# LOESS 拟合（基于清洁日）
y_loess_clean = loess(x_clean, y_clean, x_clean, frac=0.236)

# 在全量日期上做 LOESS（用清洁日拟合参数插值到所有日期）
y_loess_all = loess(x_clean, y_clean, x_all, frac=0.236)

# ---- 星期因子修正 ----
# 基于清洁日计算每周每日的平均注册量与全局均值的比值
weekday_sums = defaultdict(list)
for i in clean_idx:
    d = date.fromisoformat(dates[i])
    wd = d.weekday()  # 0=Mon, 6=Sun
    weekday_sums[wd].append(y_all[i])

global_mean_clean = y_clean.mean()
weekday_factors = {}
for wd in range(7):
    if weekday_sums[wd]:
        wd_mean = np.mean(weekday_sums[wd])
        weekday_factors[wd] = wd_mean / global_mean_clean
    else:
        weekday_factors[wd] = 1.0

day_names = ['周一','周二','周三','周四','周五','周六','周日']
print("\n星期因子（基于清洁日）:")
for wd in range(7):
    print(f"  {day_names[wd]}: {weekday_factors[wd]:.4f}")

# ---- 拟合版每日注册 ----
fitted_daily = {}
for i, d in enumerate(dates):
    wd = date.fromisoformat(d).weekday()
    if d in activity_dates:
        # 活动日/余波日：用星期修正后的 LOESS 拟合值
        fitted_daily[d] = y_loess_all[i] * weekday_factors[wd]
    else:
        # 非活动日：保留实际值
        fitted_daily[d] = y_all[i]

# ---- 月度汇总 ----
print("\n" + "=" * 70)
print("拟合版 vs 原始版 月度新增注册用户数")
print("=" * 70)
print(f"{'月份':<10} {'原始新增':>8} {'拟合新增':>8} {'差值':>8} {'压降比例':>8}")
print("-" * 50)

monthly_original = defaultdict(int)
monthly_fitted = defaultdict(float)
for d in dates:
    mon = d[:7]
    monthly_original[mon] += data[d]
    monthly_fitted[mon] += fitted_daily[d]

for mon in sorted(monthly_original.keys()):
    orig = monthly_original[mon]
    fit = monthly_fitted[mon]
    diff = orig - fit
    pct = diff / orig * 100 if orig > 0 else 0
    print(f"{mon:<10} {orig:>8,} {fit:>8,.0f} {diff:>8,.0f} {pct:>7.1f}%")

# ---- 3/4/5月详细 ----
print("\n" + "=" * 70)
print("3/4/5月 拟合版详细")
print("=" * 70)
for mon in ['2026-03', '2026-04', '2026-05']:
    orig = monthly_original[mon]
    fit = monthly_fitted[mon]
    diff = orig - fit
    pct = diff / orig * 100
    print(f"  {mon}: 原始 {orig:,} → 拟合 {fit:,.0f}（压降 {pct:.1f}%）")

# ---- 3/4/5月每日拟合 vs 原始 ----
print("\n" + "=" * 70)
print("3/4/5月每日对比（活动日标记 *）")
print("=" * 70)
print(f"{'日期':<12} {'原始':>6} {'拟合':>8} {'是否活动':>6}")
print("-" * 40)
for d in dates:
    if '2026-03' <= d <= '2026-05-31':
        is_act = '*' if d in activity_dates else ''
        print(f"{d:<12} {data[d]:>6} {fitted_daily[d]:>8.0f} {is_act:>6}")