Ladakh TET – Foundation & Exam Strategy
“Previous‑Year Papers” – Advanced Study Notes
Audience: Aspiring Ladakh TET candidates, exam‑prep coaches, and academic researchers.
Depth: MIT‑level treatment from first principles, with rigorous reasoning, deeper theory, edge‑case analysis, and a challenging problem.
1. Introduction
The Ladakh Teacher Eligibility Test (TET) is a rigorous assessment of pedagogical knowledge and subject‑specific competence.
Previous‑year papers (PYPs) are the most reliable data source for:
- Understanding exam pattern and question‑type distribution.
- Estimating the probability of various topics appearing.
- Calibrating time‑management and scoring strategies.
In this note we treat PYPs as data sets and apply quantitative, statistical, and machine‑learning tools to extract actionable insights.
2. Why PYPs Matter – A First‑Principles View
-
Empirical Law of the Exam
- Let (S) be the set of all possible questions that the exam board can design.
- The test‑making process is a random sampling from (S) conditioned on syllabus constraints (C).
- As the number of past papers (n \to \infty), the empirical distribution (P_n) converges to the true underlying distribution (P) (by the Law of Large Numbers).
- Therefore, (P_n) is the best estimate we have for (P).
-
Marking‑Scheme & Scoring
- The exam’s scoring function (f: Q \to \mathbb{R}) maps each question (Q) to a mark contribution.
- If we know (f) (e.g., +4 for MCQ, +2 for short answer), we can compute expected scores for different strategies.
-
Time Allocation
- Let (T) be total exam time, (t_i) time per question type.
- Optimal allocation maximizes expected score subject to (\sum_i t_i \le T).
- This is a constrained optimisation problem solvable via Lagrange multipliers or dynamic programming.
3. Collecting & Organizing PYPs
| Step | Action | Tools |
|---|---|---|
| 1 | Download all official Ladakh TET papers (PDFs) from the education board website. | Browser, wget/curl, pdftotext |
| 2 | Parse PDFs into structured JSON: question_id, type, subject, mark, keywords |
Python (PyPDF2, pdfminer.six) |
| 3 | Index by year, subject, question type | SQLite, Pandas DataFrame |
| 4 | Tag questions with syllabus topics using a controlled vocabulary | NLTK, spaCy, custom ontology |
Edge case: Some years may have pilot or online versions with slightly altered question formats. Flag these separately.
4. Analytical Framework
4.1 Frequency Analysis
- Definition: For a topic (x), (f(x) = \frac{\text{# questions on }x}{\text{total questions}}).
- Interpretation: High (f(x)) → high probability of appearance.
Example Calculation
Suppose across 5 years (100 questions each), Topic A appears 40 times.
(f(A) = 40/500 = 0.08) → 8 % chance per question.
4.2 Statistical Measures
| Metric | Formula | Interpretation |
|---|---|---|
| Mean | (\mu = \frac{1}{n}\sum_i q_i) | Average marks per question type |
| Variance | (\sigma^2 = \frac{1}{n}\sum_i (q_i-\mu)^2) | Spread of marks or difficulty |
| Skewness | (\frac{\frac{1}{n}\sum_i (q_i-\mu)^3}{\sigma^3}) | Bias toward harder or easier questions |
| Kurtosis | (\frac{\frac{1}{n}\sum_i (q_i-\mu)^4}{\sigma^4} - 3) | Peakedness of distribution |
Edge case: If the board changes the marking scheme (e.g., adds negative marking), recompute all metrics.
4.3 Topic‑Mark Matrix
Define matrix (