查明学费不缺且给出0856定向缺口清单

官方目录 chsi_2027_dir.tuition_fee 已覆盖80565行543校,口径是元/年:
学硕3学制均值8061元(若为全程应≈24000),与启航按校+专业码对齐1325例
1009例全等。原计划去外部站采学费是方向错——在职研源只覆盖非全,
而全日制年费库里早已有官方值。只把5539个NULL行填上,已有值不动,
差异316例里仅26例是年费与全程混用,其余按口径冲突留证。

0856定向清单:362所开0856的院校里347所缺2027推免接收数、
只有18所有2027自命题大纲、2所有逐专业线,这三项才是该优先补的。
This commit is contained in:
Your Name committed 2026-11-01 01:39:36 +08:00
1 parent d24325ba06
commit 16f8a60bb6
5 files changed
+3908

No files matched your search

File diff suppressed because it is too large. Load diff
File diff suppressed because it is too large. Load diff
@@ -0,0 +1,17 @@
{
"结论": "学费不缺。官方研招网目录 chsi_2027_dir.tuition_fee 已覆盖 80,565 行 / 543 校,口径是元/年。",
"口径证据": [
"学硕全日制且学制3年的平均学费 8061 元;若为全程应≈24000 元(国家标准学硕 8000 元/年)",
"与启航 qihang_tuition 按「校+专业码」对齐 1,325 例,1,009 例完全相等(76.2%),且相等值都落在 8000/10000/12000 这类年费档位",
"启航自身有 460 行单位「未标注」,占其总量 22%,口径不清的一方是平台不是官方"
],
"差异归因": "316 例差异里只有 26 例是 2 倍关系(年费 vs 全程混用),其余 290 例是数值本身不同,按口径冲突单列,不判谁错,也不覆盖点对点的已有值",
"回填动作": "只把 5,539 个 tuition_year 为 NULL 的点对点行填上官方值,已有值一律不动;全程额按 学制×年费 折算并标 derived",
"对采集方向的修正": "原计划派 agent 去外部站采学费是方向性错误——在职研类源只覆盖非全,而库里官方源已有全日制年费。外部源的真实增量只剩「住宿」这一新字段与非全的高额学费交叉核验",
"0856定向缺口": {
"开0856校": "362 所 / 1880 行(含学院)",
"缺2027推免接收数": "347 校 / 1775 行",
"缺2027自命题大纲": "仅 18 校有 43 行",
"有2027逐专业线": "2 校"
}
}
@@ -0,0 +1,82 @@
# -*- coding: utf-8 -*-
"""用官方 CHSI 目录的 tuition_fee 回填点对点学费空行。
只做 NULL→值,绝不覆盖点对点已有的学费:两源同键不同值属口径冲突,要留着对账。
CHSI 该字段是元/年口径(学硕 3 学制均值 8061 元,若为全程应≈24000;
与启航「元/年」行按校+专业码对齐 76.2% 完全相等)。
全程额按 学制×年费 折算,折算值单独记 tuition_total 并标 derived。
"""
import csv
import gzip
import hashlib
import io
import json
import re
import sqlite3
import time
ZX = r'D:/择校/数据库/zexiao_2027.db'
PRE = r'D:/择校/运行暂存/入库前备份/20261101_tuition_fill_pre'
def main():
z = sqlite3.connect(ZX, timeout=600)
z.execute('PRAGMA busy_timeout=600000')
q = ("select p.rowid, p.school, p.code, c.tuition_fee, c.duration_years "
"from pd_dir_detail p join chsi_2027_dir c "
"on c.dwmc=p.school and c.zydm=substr(p.code,1,6) "
"where (p.tuition_year is null or trim(p.tuition_year)='') "
"and c.tuition_fee glob '[0-9]*'")
rows = z.execute(q).fetchall()
print('候选回填行=%d' % len(rows))
if not rows:
return
import os
os.makedirs(PRE, exist_ok=True)
cols = ['rowid', 'school', 'dept', 'year', 'code', 'major', 'tuition',
'tuition_year', 'tuition_total', 'tuition_kind']
sel = ','.join('"%s"' % c for c in cols)
old = z.execute('select rowid,%s from pd_dir_detail where rowid in (%s)' % (
sel, ','.join(str(r[0]) for r in rows))).fetchall()
gz = gzip.open(PRE + '/pd_dir_detail_tuition_rows.csv.gz', 'wb')
f = io.TextIOWrapper(gz, encoding='utf-8', newline='')
w = csv.writer(f)
w.writerow(['rowid'] + cols)
for r in old:
w.writerow(list(r))
f.close()
gz.close()
sha = hashlib.sha256(open(PRE + '/pd_dir_detail_tuition_rows.csv.gz', 'rb').read()).hexdigest()
now = time.strftime('%Y-%m-%dT%H:%M:%S')
filled = derived = 0
for rid, school, code, fee, dur in rows:
v = re.search(r'\d+', str(fee)).group()
d = None
if dur and re.search(r'\d', str(dur)):
dd = float(re.search(r'\d+(\.\d+)?', str(dur)).group())
if 1 <= dd <= 8:
d = str(int(round(float(v) * dd)))
derived += 1
z.execute('update pd_dir_detail set tuition_year=?, tuition_total=coalesce(tuition_total,?),'
' tuition_kind=? where rowid=? and (tuition_year is null or trim(tuition_year)=\'\')',
(v, d, '官方CHSI2027(元/年)', rid))
filled += 1
z.execute('INSERT INTO changelog VALUES (?,?,?,?,?,?)',
(time.strftime('%Y-%m-%d %H:%M:%S'), 'tuition_fill_20261101',
'pd_dir_detail', 'update_null_to_value', filled,
'用官方CHSI tuition_fee 回填学费NULL;全程额按学制折算%d行(标derived);'
'备份 %s' % (derived, sha[:16])))
z.commit()
json.dump({'table': 'pd_dir_detail', 'rows_targeted': len(rows), 'rows_filled': filled,
'derived_total': derived, 'backup_sha256': sha, 'at': now},
open(PRE + '/receipt.json', 'w', encoding='utf-8'), ensure_ascii=False, indent=1)
left = z.execute("select count(*) from pd_dir_detail where tuition_year is null or trim(tuition_year)=''").fetchone()[0]
tot = z.execute("select count(*) from pd_dir_detail where tuition_year glob '[0-9]*'").fetchone()[0]
print('回填=%d(其中折算全程=%d) 学费空行余=%d 有值行=%d' % (filled, derived, left, tot))
print('quick_check=', z.execute('PRAGMA quick_check').fetchone()[0])
z.close()
if __name__ == '__main__':
main()
+47
View File
@@ -0,0 +1,47 @@
# -*- coding: utf-8 -*-
"""量化官方 CHSI 学费字段能补多少点对点空行,并归因两源差异。"""
import re
import sqlite3
z = sqlite3.connect(r'D:/择校/数据库/zexiao_2027.db', timeout=260)
cols = [c[0] for c in z.execute('select * from pd_dir_detail limit 0').description]
tc = [c for c in cols if 'tuition' in c.lower()]
print('学费相关列:', tc)
if tc:
t0 = tc[0]
q = ("select count(*) tot, sum(case when c.tuition_fee is not null and trim(c.tuition_fee)<>'' "
"then 1 else 0 end) can from pd_dir_detail p left join chsi_2027_dir c "
"on c.dwmc=p.school and c.zydm=substr(p.code,1,6) "
"where p.\"%s\" is null or trim(p.\"%s\")=''" % (t0, t0))
print('点对点学费空行 / 其中 CHSI 有官方值:', z.execute(q).fetchone())
rows = z.execute(
"select c.tuition_fee,q.tuition_amount,q.tuition_unit,c.xxfsmc "
"from chsi_2027_dir c join qihang_tuition q "
"on q.school=c.dwmc and substr(q.major_code,1,6)=c.zydm "
"where c.tuition_fee glob '[0-9]*' and q.tuition_amount glob '[0-9]*' limit 8000").fetchall()
twice = other = same = 0
unit = {}
half = 0
for chi, qih, u, xxfs in rows:
a, b = int(re.search(r'\d+', str(chi)).group()), int(re.search(r'\d+', str(qih)).group())
if a == b:
same += 1
continue
if b == 2 * a or a == 2 * b:
twice += 1
else:
other += 1
if b > a:
half += 1
k = str(u)[:10]
unit[k] = unit.get(k, 0) + 1
print('可比=%d 相等=%d(%.1f%%) 差异=%d(其中2倍关系=%d,一方高于另一方=%d)'
% (len(rows), same, 100.0 * same / max(1, len(rows)), twice + other, twice, half))
print('差异例的启航单位分布:', unit)
print('全体启航单位:', z.execute(
"select tuition_unit,count(*) from qihang_tuition group by 1 order by 2 desc limit 5").fetchall())
print('\nCHSI 官方学费覆盖: 行=%s 校=%s' % z.execute(
"select count(*),count(distinct dwdm) from chsi_2027_dir "
"where tuition_fee glob '[0-9]*'").fetchone())