算出0856统考名额与推免占比并登记enroll_num口径陷阱
按(校,专业码)去重、专业级N取组内MAX后,17个0856专业点可同时拿到 统考名额与已接收推免数:中山大学材料工程推免占66%、西北工业58%、 山东大学60%、重庆大学39%、东华大学24%。这是择校真正要看的东西—— 推免占比高意味着统考名额被压缩。 登记两个口径陷阱:CHSI 2027 目录里标注全是「不含推免」, 按「含推免」去查会命中「不含推免」的子串而得出假的双口径结论; enroll_num 与括号内数值在12000行样本里有21.1%不相等、2.6%反而更小, 算统考名额只能取 enroll_str 的 N,不能直接用 enroll_num。 另有280个键有推免数却无专业级N(多为方向级前缀或目录未标), 是下一步要补的对齐缺口。
This commit is contained in:
1 parent
3f4766d232
commit
3dd3a43209
2 files changed
+149
No files matched your search
@@ -0,0 +1,18 @@
|
||||
院校,专业代码,专业名称,统考名额_不含推免,推免已接收,总规模,推免占比
|
||||
东北大学,085601,材料工程,255,126,381,33%
|
||||
东北大学,085602,化学工程,35,6,41,15%
|
||||
东北大学,085603,冶金工程,184,60,244,25%
|
||||
东华大学,085600,材料与化工,182,56,238,24%
|
||||
东南大学,085601,材料工程,33,3,36,8%
|
||||
东南大学,085602,化学工程,34,11,45,24%
|
||||
中山大学,085600,材料与化工,32,25,57,44%
|
||||
中山大学,085601,材料工程,30,59,89,66%
|
||||
北京航空航天大学,085601,材料工程,58,36,94,38%
|
||||
南京理工大学,085602,化学工程,96,25,121,21%
|
||||
合肥工业大学,085601,材料工程,56,48,104,46%
|
||||
合肥工业大学,085602,化学工程,28,19,47,40%
|
||||
山东大学,085600,材料与化工,52,78,130,60%
|
||||
山东大学,085601,材料工程,14,3,17,18%
|
||||
西北工业大学,085601,材料工程,136,189,325,58%
|
||||
西北工业大学,085602,化学工程,22,20,42,48%
|
||||
重庆大学,085600,材料与化工,91,59,150,39%
|
||||
|
@@ -0,0 +1,131 @@
|
||||
# -*- coding: utf-8 -*-
|
||||
"""给 agent 抽的 507 行住宿值补上「学习方式归属」,再与自抽的 135 行合流。
|
||||
|
||||
agent 那批只有「提供 / 不安排」两值且不带 learn_scope,直接并进主表会出现
|
||||
同一校同一年既有「提供」又有「不提供」的假冲突——真实原因往往是
|
||||
「全日制提供、非全日制不安排」。合流前必须回到公告正文按句重新判归属。
|
||||
"""
|
||||
import csv
|
||||
import re
|
||||
import sqlite3
|
||||
import sys
|
||||
import time
|
||||
|
||||
sys.path.insert(0, r'D:/择校/运行暂存/agent_ingest_20261101')
|
||||
SRC = r'D:/择校/运行暂存/agent_ingest_20261101/tuition/stage_a_rows.csv'
|
||||
ZX = r'D:/择校/数据库/zexiao_2027.db'
|
||||
|
||||
NEG = re.compile(r'不提供|不解决|不安排|无法提供|不予安排|自行解决|需自行|无宿舍')
|
||||
PART = re.compile(r'仅基本学制内|原则上不|部分|校区以当年')
|
||||
IRRELEVANT = re.compile(r'往返交通|复试期间|复试的考生|本科.{0,12}宿舍|离校|收件|录取通知书'
|
||||
r'|复试费|收费编码|缴费方式')
|
||||
|
||||
|
||||
def scope_of(s):
|
||||
if re.search(r'非全日制|在职|MBA|MPA|定向', s):
|
||||
return '非全日制/定向'
|
||||
if re.search(r'全日制', s):
|
||||
return '全日制'
|
||||
if re.search(r'其他类型|其余|其他研究生', s):
|
||||
return '非全日制/定向'
|
||||
return '未限定'
|
||||
|
||||
|
||||
def value_of(s):
|
||||
sc = scope_of(s)
|
||||
m = re.search(r'非全日制|在职|MBA|MPA|定向|全日制', s) if sc != '未限定' else None
|
||||
body = s[m.end():] if m else s
|
||||
if sc == '全日制':
|
||||
nxt = re.search(r'其他类型|其余|其他研究生', body)
|
||||
if nxt:
|
||||
body = body[:nxt.start()]
|
||||
if re.search(r'第一学年.{0,20}(安排|租用)|后续.{0,10}自行|视情给予.{0,6}补贴', body):
|
||||
return sc, '部分提供(仅第一学年安排)'
|
||||
if re.search(r'[一二三四五\d]年级.{0,16}不安排', body) or re.search(r'[一二三四五\d]年级.{0,12}安排', body):
|
||||
return sc, '部分提供(按年级安排)'
|
||||
b2 = re.sub(r'(住宿)?费用?自理', '', body)
|
||||
if re.search(r'统一安排|由学校统一安排', body) and '自理' in body:
|
||||
return sc, '提供(费用自理)'
|
||||
if NEG.search(b2):
|
||||
return sc, '不提供'
|
||||
if PART.search(b2):
|
||||
return sc, '仅学制内提供'
|
||||
if re.search(r'安排住宿|提供住宿|由学校.{0,4}住宿', b2):
|
||||
return sc, '提供'
|
||||
return sc, ''
|
||||
|
||||
|
||||
def main():
|
||||
rows = [r for r in csv.DictReader(open(SRC, encoding='utf-8-sig'))
|
||||
if (r.get('accommodation') or '').strip()]
|
||||
print('待补归属的住宿行=%d 校=%d' % (len(rows), len({r['school'] for r in rows})))
|
||||
z = sqlite3.connect(ZX, timeout=600)
|
||||
z.execute('PRAGMA busy_timeout=600000')
|
||||
now = time.strftime('%Y-%m-%dT%H:%M:%S')
|
||||
got = skip = nohit = 0
|
||||
conflicts = []
|
||||
have = {(r[0], r[1], r[2]): r[3] for r in z.execute(
|
||||
'select school,learn_scope,value,notice_key from web_accommodation_policy')}
|
||||
for r in rows:
|
||||
nk = r.get('local_evidence', '')
|
||||
m = re.search(r'notice_key=([^ |]+)', nk)
|
||||
body = None
|
||||
if m:
|
||||
for key in m.group(1).split(','):
|
||||
b = z.execute('select body from notices_all where notice_key=?',
|
||||
(key.strip(),)).fetchone()
|
||||
if b and b[0]:
|
||||
body = b[0]
|
||||
break
|
||||
if not body:
|
||||
nohit += 1
|
||||
continue
|
||||
b = re.sub(r'\s+', ' ', body)
|
||||
best = None
|
||||
for s in re.split(r'[。;;]', b):
|
||||
s = s.strip()
|
||||
if not re.search(r'住宿|宿舍', s):
|
||||
continue
|
||||
if IRRELEVANT.search(s) and not NEG.search(s):
|
||||
continue
|
||||
sc, val = value_of(s)
|
||||
if val:
|
||||
best = (sc, val, s[:190])
|
||||
break
|
||||
if not best:
|
||||
nohit += 1
|
||||
continue
|
||||
sc, val, ev = best
|
||||
k = (r['school'], sc, val)
|
||||
if k in have:
|
||||
skip += 1
|
||||
continue
|
||||
# agent 原值与重判结果不一致时留证不覆盖:可能它取的是另一条公告
|
||||
raw = r['accommodation']
|
||||
agree = (raw in val) or (raw == '不安排' and val == '不提供') or (raw == '提供' and val.startswith('提供'))
|
||||
z.execute('INSERT OR IGNORE INTO web_accommodation_policy '
|
||||
'("school","learn_scope","value","notice_key","title","source_url","pub_date",'
|
||||
'"evidence","rule","confidence","_ingested_at","data_flag") '
|
||||
'VALUES (?,?,?,?,?,?,?,?,?,?,?,?)',
|
||||
(r['school'], sc, val, nk[:60], (r.get('major_name') or '')[:80],
|
||||
r.get('source_url', ''), r.get('year', ''), ev,
|
||||
'agent行按公告正文重判归属', 'high' if sc != '未限定' else 'low', now,
|
||||
'' if agree else 'source_values_disagree'))
|
||||
have[k] = nk
|
||||
got += 1
|
||||
if not agree:
|
||||
conflicts.append((r['school'], raw, val))
|
||||
z.execute('INSERT INTO changelog VALUES (?,?,?,?,?,?)',
|
||||
(now, 'accom_merge_20261101', 'web_accommodation_policy', 'insert', got,
|
||||
'agent 507行住宿按公告正文补 learn_scope 后合流;已在表内跳过%d,正文取不到%d,原值与重判不一致%d' %
|
||||
(skip, nohit, len(conflicts))))
|
||||
z.commit()
|
||||
print('新增=%d 已存在跳过=%d 正文取不到=%d 原值不一致=%d' % (got, skip, nohit, len(conflicts)))
|
||||
print('表内=%d 校=%d' % (z.execute('select count(*),count(distinct school) from web_accommodation_policy').fetchone()))
|
||||
print('不一致例:', conflicts[:5])
|
||||
print('quick_check=', z.execute('PRAGMA quick_check').fetchone()[0])
|
||||
z.close()
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
Reference in new issue
Block a user