← All projects · Case study
Lilawati Vidya Mandir: Class 10 & 12 pre-mid-term, graded end to end
The school sent its question papers, marking schemes and every handwritten answer sheet for Class 10 and Class 12. EvalAssist AI read each sheet, marked every question against the school's own scheme, and turned the results into class, section, chapter and skill analysis. Student names are not shown on this page.
The hand-off: what the school gave us
A library of 65 documents: 23 question papers, 23 marking schemes and 18 sample answer sheets, plus the scanned bundles of answer sheets for every section. Each document is fingerprinted, so a file sent twice is caught as a duplicate.
- Class 10 Biology: no answer-sheet sample
- Class 10 French: no marking scheme, no answer-sheet sample
- Class 12 Biology: no answer-sheet sample
- Class 12 Business Studies: no answer-sheet sample
- Class 12 Economics: no answer-sheet sample
- Class 12 Sociology: no marking scheme
Results by subject
Overall average 43% (Class 10: 44%, Class 12: 41%). This was a pre-mid-term exam, so it shows where teaching time should go next, not final board readiness. Sorted weakest first.
| Subject | Sheets | Average | Median | Below 33% |
|---|---|---|---|---|
| Mathematics | 165 | 30% | 28% | 64% |
| Social Science | 173 | 39% | 35% | 48% |
| Hindi | 166 | 44% | 42% | 30% |
| Biology | 98 | 46% | 43% | 33% |
| Science | 168 | 48% | 44% | 30% |
| French | 16 | 58% | 62% | 12% |
| English | 180 | 58% | 58% | 12% |
| Subject | Sheets | Average | Median | Below 33% |
|---|---|---|---|---|
| Informatics Practices | 87 | 19% | 15% | 80% |
| Economics | 66 | 31% | 29% | 65% |
| Physics | 59 | 37% | 31% | 51% |
| English | 153 | 40% | 38% | 37% |
| History | 21 | 41% | 40% | 38% |
| Mathematics | 75 | 42% | 40% | 35% |
| Accountancy | 67 | 44% | 45% | 25% |
| Chemistry | 58 | 45% | 43% | 31% |
| Business Administration | 15 | 50% | 50% | 7% |
| Physical Education | 24 | 52% | 51% | 4% |
| Political Science | 22 | 53% | 55% | 14% |
| Business Studies | 69 | 53% | 52% | 19% |
| Marketing | 13 | 53% | 49% | 8% |
| Biology | 20 | 54% | 57% | 20% |
| Sociology | 41 | 60% | 60% | 10% |
Section × subject heatmap
Average score per section, per subject (subjects with at least 20 sheets; cells with under 3 sheets are hidden). Hover a cell for the exact figure. This is the view a principal uses to see which section needs help in which subject.
| Section | Mathematics | Social Science | Hindi | Biology | Science | English |
|---|---|---|---|---|---|---|
| 10-A | 35% | 63% | 52% | — | 56% | 68% |
| 10-B | 23% | 55% | 39% | — | 39% | 61% |
| 10-C | 29% | 27% | 47% | 48% | 52% | 56% |
| 10-D | 34% | 39% | 47% | 52% | 51% | 62% |
| 10-E | 34% | 28% | 41% | 42% | 47% | 55% |
| 10-F | 34% | 30% | 41% | 54% | 48% | 58% |
| 10-G | 23% | 38% | 40% | 40% | 42% | 50% |
| All | 30% | 39% | 44% | 46% | 48% | 58% |
| Section | Informatics Practices | Economics | Physics | English | History | Mathematics | Accountancy | Chemistry | Physical Education | Political Science | Business Studies | Biology | Sociology |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 12-A | 23% | — | — | 35% | 41% | — | — | — | — | 53% | — | — | 59% |
| 12-B | 24% | — | — | 22% | — | — | — | — | — | — | — | — | 61% |
| 12-C | — | 43% | — | 52% | — | 32% | 52% | — | — | — | 62% | — | — |
| 12-D | 13% | 22% | — | 34% | — | 23% | 38% | — | — | — | 48% | — | — |
| 12-E | — | — | 44% | 51% | — | 57% | — | 51% | — | — | — | — | — |
| 12-F | — | — | 32% | 44% | — | 42% | — | 40% | 52% | — | — | 54% | — |
| All | 19% | 31% | 37% | 40% | 41% | 42% | 44% | 45% | 52% | 53% | 53% | 54% | 60% |
Where the marks were lost
Every question was tagged with its thinking level and the skill it tests. The share of available marks students actually scored:
By thinking level (Bloom)
By skill
Weakest chapters (at least 25 questions)
| Class | Subject | Chapter | Scored | Questions |
|---|---|---|---|---|
| Class 12 | Economics | Rural Development | 8% | 65 |
| Class 12 | Economics | Liberalisation, Privatisation and Globalisation: An Appraisal | 14% | 57 |
| Class 12 | Economics | Indian Economy 1950-1990 | 17% | 125 |
| Class 12 | Economics | Human Capital Formation in India | 18% | 66 |
| Class 12 | Informatics Practices | Societal Impacts | 19% | 58 |
| Class 12 | Informatics Practices | Structured Query Language | 21% | 2245 |
| Class 12 | English | Reading Comprehension | 22% | 975 |
| Class 10 | Mathematics | Areas Related to Circles | 23% | 1142 |
| Class 12 | Economics | National Income Accounting | 27% | 489 |
| Class 12 | History | Thinkers, Beliefs and Buildings: Cultural Developments | 29% | 107 |
| Class 10 | Mathematics | Pair of Linear Equations in Two Variables | 30% | 1144 |
| Class 12 | History | Kings, Farmers and Towns: Early States and Economies | 32% | 72 |
How the grading ran, and the checks on it
Sheets are flagged when the handwriting is hard to read, a diagram breaks the question split, the scheme is missing, or the marks look inconsistent with the answer. A teacher confirms or corrects those marks before reports are released.
People checking the machine
A team of reviewers checked how well the system read the question papers and marking schemes: 1,762 questions across 64 of 65 documents, side by side with the original scan.
Reading quality
What the reviewers flagged
What we learned and fixed
Names on cover sheets are an observation, not an identity. In one section, a few handwritten names were misread (for example, one first name read as a similar-looking one with the same surname). The sheet count was right; the names were not.
The fix: every sheet is now matched against the school's official class list. Only a unique, confirmed match is published; a conflicting name or roll number holds the sheet for review.
Next steps with the school
- Teacher review of the flagged sheets, then reports released to students and parents.
- Section-wise remediation plans for the weakest chapters above.
- The same analysis after the mid-term, to measure the change section by section.
Want this for your school? Talk to us · See the Punjab & Puducherry course-run study
Data: EvalAssist AI production records for Lilawati Vidya Mandir, graded 14–17 Jul 2026 (duplicates excluded); reviewer data to 4 Aug 2026. Figures are system marks before teacher review.