# 5. Target Exam Analytics

User Type: **Super Administrator**
Source: *Mi Digital Academy - Education CRM Features Document*

---

## 5. Target Exam Analytics

### 5.1 Prediction Accuracy Analytics
**What it does:** Analyzes the accuracy of AI score predictions against actual exam results.
**Sub-features:**
- Predicted vs actual score comparison (per student, per exam).
- Mean absolute error (MAE) and bias (over/under prediction).
- Accuracy by exam type and cohort.
- Calibration analysis (predicted confidence vs observed accuracy).
- available on web
- event logging (action performed)
- Audit logging of prediction accuracy analytics
**Super Admin User Journey:**
1. Open Target Exam Management → Analytics → Prediction Accuracy.
2. Review the predicted vs actual score comparison.
3. Inspect the MAE and bias metrics.
4. Slice the accuracy by exam type and cohort.
5. Review the calibration analysis.
**Rules & Edge Cases:**
- A student with no actual result is excluded from accuracy metrics.
- A bias metric of zero indicates no systematic over/under prediction.
- A cohort with fewer than the minimum completed exams shows "insufficient data".
- An accuracy metric is never estimated for missing actuals.

### 5.2 Plan Adherence Analytics
**What it does:** Analyzes how well students adhere to their generated study plans.
**Sub-features:**
- Adherence rate (on-time task completion percentage).
- Slippage analysis (average days tasks slip).
- Adherence by exam type, cohort, and plan horizon.
- At-risk and off-track student identification.
- available on web
- event logging (action performed)
- Audit logging of plan adherence analytics
**Super Admin User Journey:**
1. Open Analytics → Plan Adherence.
2. Review the adherence rate.
3. Inspect the slippage analysis.
4. Slice adherence by exam type, cohort, and horizon.
5. Identify at-risk and off-track students.
**Rules & Edge Cases:**
- A plan with no completed tasks shows 0% adherence (not excluded).
- A slippage metric is computed only over completed tasks.
- An at-risk student is one at the at-risk threshold (not yet off-track).
- A cohort adherence with no active plans shows "no data".

### 5.3 Mock Test Trend Analytics
**What it does:** Analyzes mock test score progression over time.
**Sub-features:**
- Score trend (per student across mocks).
- Improvement rate (score delta between mocks).
- Trend by exam type and cohort.
- Plateau detection (no improvement over N mocks).
- available on web
- event logging (action performed)
- Audit logging of mock test trend analytics
**Super Admin User Journey:**
1. Open Analytics → Mock Test Trends.
2. Review the score trend per student.
3. Inspect the improvement rate.
4. Slice the trend by exam type and cohort.
5. Identify plateauing students.
**Rules & Edge Cases:**
- A student with fewer than two mocks shows no trend (insufficient data).
- A plateau is detected only after the configured number of mocks (N).
- A score delta is computed on comparable mocks (same exam type).
- A cohort trend with no mock data shows "no data".

### 5.4 Gap Closure Analytics
**What it does:** Analyzes the closure of identified knowledge gaps over time.
**Sub-features:**
- Gap identification (weak topics from mocks and practice).
- Gap closure rate (weak topics improved to target).
- Gap persistence (topics remaining weak over time).
- Gap closure by exam type and cohort.
- available on web
- event logging (action performed)
- Audit logging of gap closure analytics
**Super Admin User Journey:**
1. Open Analytics → Gap Closure.
2. Review the gap identification (weak topics).
3. Inspect the gap closure rate.
4. Check the gap persistence.
5. Slice the closure by exam type and cohort.
**Rules & Edge Cases:**
- A gap is closed only when the topic accuracy reaches the target threshold.
- A persistent gap (weak over the configured period) is flagged for intervention.
- A gap closure rate with no identified gaps shows "no gaps".
- A topic with insufficient attempts is not counted as closed or persistent.

### 5.5 Cohort & Institution-Level Reporting
**What it does:** Produces cohort and institution-level reports for the target exam feature.
**Sub-features:**
- Cohort report (aggregated prediction, adherence, mock, gap metrics).
- Institution report (per institution, comparable cohorts).
- Report export (PDF, CSV).
- Report scheduling (periodic delivery).
- available on web
- event logging (action performed)
- Audit logging of cohort & institution-level reporting
**Super Admin User Journey:**
1. Open Analytics → Reports.
2. Generate a cohort report.
3. Generate an institution report with comparable cohorts.
4. Export the report (PDF, CSV).
5. Schedule periodic report delivery.
**Rules & Edge Cases:**
- A cohort report with no students shows "no data" (not an empty chart).
- An institution report compares only cohorts with comparable exam types.
- A report export with no data exports a header-only file (not an error).
- A scheduled report delivery respects the recipient's email preferences.