ca3780aa24
包含 CDC consumer、analytics/mastery/warning service、grpc server、repository、ClickHouse DDL 等
132 lines
6.5 KiB
SQL
132 lines
6.5 KiB
SQL
-- data-ana 模块 Mock 数据脚本
|
||
-- 用途：本地开发/集成测试时为 5 张宽表插入演示数据，验证查询逻辑
|
||
-- 使用方式（需先执行 clickhouse-init.sql）：
|
||
-- clickhouse-client --multiquery < scripts/data-ana-mock-data.sql
|
||
--
|
||
-- 数据规模：2 班 × 10 学生 × 3 知识点 × 2 考试 = 120 行 student_dashboard_view
|
||
-- 20 条错题 / 60 条掌握度快照 / 20 条考勤 / 10 条 AI 用量
|
||
-- 所有时间戳使用 now() - N days 模式，避免硬编码绝对日期导致脚本过期
|
||
|
||
USE edu_analytics;
|
||
|
||
-- ===== 1. student_dashboard_view：学情宽表 =====
|
||
-- 班级 C001（高一 1 班）+ C002（高一 2 班），每班 10 学生
|
||
-- 2 次考试（EXAM001 数学月考 / EXAM002 数学期中），3 个数学知识点
|
||
INSERT INTO student_dashboard_view
|
||
SELECT
|
||
'S' || toString(number + 1) AS student_id, -- S001..S020
|
||
IF(number < 10, 'C001', 'C002') AS class_id, -- 前 10 在 C001，后 10 在 C002
|
||
'EXAM001' AS exam_id,
|
||
'SUBJ_MATH' AS subject_id,
|
||
40 + (number % 60) AS score, -- 40-99 分
|
||
(number % 10) + 1 AS rank_in_class, -- 1-10 名
|
||
'KP_MATH_' || toString((number % 3) + 1) AS knowledge_point_id, -- KP_MATH_1/2/3
|
||
0.3 + ((number % 70) / 100.0) AS mastery_level, -- 0.30-0.99
|
||
(number % 5) AS error_count, -- 0-4 错题
|
||
now() - INTERVAL 7 DAY AS last_updated
|
||
FROM numbers(20);
|
||
|
||
-- 第二次考试（EXAM002），成绩略升
|
||
INSERT INTO student_dashboard_view
|
||
SELECT
|
||
'S' || toString(number + 1) AS student_id,
|
||
IF(number < 10, 'C001', 'C002') AS class_id,
|
||
'EXAM002' AS exam_id,
|
||
'SUBJ_MATH' AS subject_id,
|
||
50 + (number % 50) AS score, -- 50-99 分
|
||
(number % 10) + 1 AS rank_in_class,
|
||
'KP_MATH_' || toString((number % 3) + 1) AS knowledge_point_id,
|
||
0.4 + ((number % 60) / 100.0) AS mastery_level,
|
||
(number % 4) AS error_count,
|
||
now() - INTERVAL 1 DAY AS last_updated
|
||
FROM numbers(20);
|
||
|
||
-- ===== 2. student_errors：错题本 =====
|
||
INSERT INTO student_errors
|
||
SELECT
|
||
'S' || toString((number % 20) + 1) AS student_id,
|
||
'Q_' || toString(number + 1) AS question_id,
|
||
'KP_MATH_' || toString((number % 3) + 1) AS knowledge_point_id,
|
||
(number % 5) + 1 AS error_count,
|
||
now() - INTERVAL (number % 30) DAY AS last_error_time,
|
||
'错误类型：' || multiIf(
|
||
number % 3 = 0, '计算失误',
|
||
number % 3 = 1, '公式记忆错误',
|
||
'审题不清'
|
||
) AS content
|
||
FROM numbers(20);
|
||
|
||
-- ===== 3. mastery_snapshot：掌握度历史快照（3 次计算，便于趋势查询） =====
|
||
INSERT INTO mastery_snapshot
|
||
SELECT
|
||
'S' || toString((number % 20) + 1) AS student_id,
|
||
'KP_MATH_' || toString((number % 3) + 1) AS knowledge_point_id,
|
||
'SUBJ_MATH' AS subject_id,
|
||
0.3 + ((number % 70) / 100.0) AS mastery_level,
|
||
now() - INTERVAL 30 DAY AS calculated_at,
|
||
'weighted_moving_avg' AS calculation_method
|
||
FROM numbers(20);
|
||
|
||
INSERT INTO mastery_snapshot
|
||
SELECT
|
||
'S' || toString((number % 20) + 1) AS student_id,
|
||
'KP_MATH_' || toString((number % 3) + 1) AS knowledge_point_id,
|
||
'SUBJ_MATH' AS subject_id,
|
||
0.4 + ((number % 60) / 100.0) AS mastery_level,
|
||
now() - INTERVAL 15 DAY AS calculated_at,
|
||
'weighted_moving_avg' AS calculation_method
|
||
FROM numbers(20);
|
||
|
||
INSERT INTO mastery_snapshot
|
||
SELECT
|
||
'S' || toString((number % 20) + 1) AS student_id,
|
||
'KP_MATH_' || toString((number % 3) + 1) AS knowledge_point_id,
|
||
'SUBJ_MATH' AS subject_id,
|
||
0.5 + ((number % 50) / 100.0) AS mastery_level,
|
||
now() AS calculated_at,
|
||
'weighted_moving_avg' AS calculation_method
|
||
FROM numbers(20);
|
||
|
||
-- ===== 4. attendance_logs：考勤记录（近 5 天，含缺勤/迟到） =====
|
||
INSERT INTO attendance_logs
|
||
SELECT
|
||
'S' || toString((number % 20) + 1) AS student_id,
|
||
IF(number % 20 < 10, 'C001', 'C002') AS class_id,
|
||
today() - (number % 5) AS attendance_date,
|
||
multiIf(
|
||
number % 10 = 0, 'absent', -- 10% 缺勤
|
||
number % 5 = 0, 'late', -- 20% 迟到
|
||
'present' -- 其余正常
|
||
) AS status,
|
||
'T001' AS recorded_by,
|
||
multiIf(number % 10 = 0, '病假', '') AS remark,
|
||
now() - INTERVAL (number % 5) DAY AS occurred_at
|
||
FROM numbers(20);
|
||
|
||
-- ===== 5. ai_usage_log：AI 用量计费（模拟 3 个 provider） =====
|
||
INSERT INTO ai_usage_log
|
||
SELECT
|
||
'REQ_' || toString(number + 1) AS request_id,
|
||
'S' || toString((number % 20) + 1) AS user_id,
|
||
arrayElement(['openai', 'anthropic', 'baichuan'], (number % 3) + 1) AS provider,
|
||
arrayElement(['gpt-4o-mini', 'claude-3-haiku', 'baichuan2-turbo'], (number % 3) + 1) AS model,
|
||
100 + (number % 200) AS prompt_tokens, -- 100-299
|
||
50 + (number % 150) AS completion_tokens, -- 50-199
|
||
150 + (number % 350) AS total_tokens, -- 150-499
|
||
500 + (number % 2000) AS latency_ms, -- 500-2499 ms
|
||
number % 5 != 0 AS success, -- 80% 成功
|
||
(number % 50) + 1 AS cost_cents, -- 1-50 分
|
||
now() - INTERVAL (number % 24) HOUR AS occurred_at
|
||
FROM numbers(20);
|
||
|
||
-- ===== 校验查询（可选执行） =====
|
||
SELECT 'student_dashboard_view' AS table_name, count() AS cnt FROM student_dashboard_view FINAL
|
||
UNION ALL
|
||
SELECT 'student_errors', count() FROM student_errors FINAL
|
||
UNION ALL
|
||
SELECT 'mastery_snapshot', count() FROM mastery_snapshot
|
||
UNION ALL
|
||
SELECT 'attendance_logs', count() FROM attendance_logs FINAL
|
||
UNION ALL
|
||
SELECT 'ai_usage_log', count() FROM ai_usage_log FINAL;
|