Chapter map

1. 1. Criterion
2. 2. Two clusters
3. 3. Three clusters
4. 4. Four clusters
5. 5. Five clusters
6. 6. Raw sequence
7. 7. First difference one
8. 8. First difference two
9. 9. First difference three
10. 10. Second difference at three
11. 11. Second difference at four
12. 12. Software selection
13. 13. Historical minimum
14. 14. Singularity
15. 15. Candidate range
16. 16. Claim

Visit the How Mathematics Works hub for the wider series and its connections to education, reasoning, science, computing, design and careers.

---

1. Criterion

The useful question is not only how to calculate, but what the calculation assumes. Define Mk=k^2 det(Wk).

A concrete situation

Penalty and generalised dispersion are explicit.

What the mathematics does not license

Spell Marriott consistently even if software says marriot. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

State formula. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

2. Two clusters

A small example makes the mechanism visible. Set det(W2)=25.

A concrete situation

M2=4×25=100.

What the mathematics does not license

Use squared k, not k. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Show step. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

3. Three clusters

This is where modelling choices begin to affect evidence. Set det(W3)=70/9.

A concrete situation

M3=9×70/9=70.

What the mathematics does not license

Retain exact fractions where useful. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Show step. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

4. Four clusters

The central mathematical idea is concrete. Set det(W4)=3.75.

A concrete situation

M4=16×3.75=60.

What the mathematics does not license

Same scatter convention is required. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Show step. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

5. Five clusters

Here the geometry becomes operational. Set det(W5)=2.32.

A concrete situation

M5=25×2.32=58.

What the mathematics does not license

A smaller determinant is expected with larger k. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Show step. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

6. Raw sequence

This step deserves a careful pause. List 100,70,60,58.

A concrete situation

Decline slows after k=3.

What the mathematics does not license

Do not choose terminal k automatically. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Plot sequence. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

7. First difference one

The useful question is not only how to calculate, but what the calculation assumes. Compute 70-100.

A concrete situation

Delta3=-30.

What the mathematics does not license

Index differences need an order convention. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Record value. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

8. First difference two

A small example makes the mechanism visible. Compute 60-70.

A concrete situation

Delta4=-10.

What the mathematics does not license

Negative means the criterion fell. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Record value. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

9. First difference three

This is where modelling choices begin to affect evidence. Compute 58-60.

A concrete situation

Delta5=-2.

What the mathematics does not license

The sequence is flattening. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Record value. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

10. Second difference at three

The central mathematical idea is concrete. Compute 100-2×70+60.

A concrete situation

D2 at k=3 is 20.

What the mathematics does not license

Centre the label correctly. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Show step. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

11. Second difference at four

Here the geometry becomes operational. Compute 70-2×60+58.

A concrete situation

D2 at k=4 is 8.

What the mathematics does not license

Use identical rounding. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Show step. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

12. Software selection

This step deserves a careful pause. Compare 20 and 8.

A concrete situation

Maximum second difference marks k=3.

What the mathematics does not license

This is a declared convention. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

State result. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

13. Historical minimum

The useful question is not only how to calculate, but what the calculation assumes. Inspect k^2 det(Wk) directly.

A concrete situation

A direct-minimum rule can disagree on a finite range.

What the mathematics does not license

Do not merge conventions. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Report both. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

14. Singularity

A small example makes the mechanism visible. Let W lose rank.

A concrete situation

det(W)=0 can dominate the score.

What the mathematics does not license

Regularisation changes the criterion. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Flag failure. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

15. Candidate range

This is where modelling choices begin to affect evidence. Extend k beyond five.

A concrete situation

New curvature or minima may appear.

What the mathematics does not license

Range is part of the evidence. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Run sensitivity. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

16. Claim

The central mathematical idea is concrete. Name W,k range,penalty and difference rule.

A concrete situation

Teaching curvature supports k=3.

What the mathematics does not license

Do not promise natural groups. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Bound conclusion. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

---

A seven-day practice lab

1. Day 1: define the decision, observation unit, X variables and Y or reference condition.
2. Day 2: reproduce the smallest worked example and check every dimension, unit and sign.
3. Day 3: build a transparent baseline and write what it assumes.
4. Day 4: fit the main method inside a training-only pipeline; record every parameter.
5. Day 5: challenge one assumption with a deliberately difficult dataset or operating regime.
6. Day 6: compare validation metrics with the practical workload created by errors.
7. Day 7: present one supported conclusion, one unresolved uncertainty and one safe next experiment.

Questions for a student, parent or mentor

• Which part of the model is a mathematical definition, and which part is a choice?
• What small calculation would catch a sign, scale or dimension error?
• What evidence comes from training, validation and the untouched test?
• What benign condition could look like a fault or prediction failure?
• How would the answer change for another response, asset, school project or time period?
• What should a human review before acting on the score?

A method-specific audit before deployment

The Marriott index begins with determinant Wk, the generalised within-cluster dispersion, and multiplies it by k squared. The determinant normally falls as more clusters are allowed, while k squared pushes back against needless fragmentation. Marriott’s original discussion uses k squared times determinant Wk as a practical cluster-count device; widely used software may locate a pronounced bend through second differences. Those rules are related but not interchangeable, so the article keeps the raw index and the selection rule on separate lines.

Suppose the raw Marriott scores k squared times determinant Wk are 100 at k=2, 70 at k=3, 60 at k=4 and 58 at k=5. These values correspond to determinants 25, 70/9 or about 7.7778, 3.75 and 2.32. The determinant keeps shrinking, yet the penalised score soon flattens.

The first differences are minus 30, minus 10 and minus 2. The centred second difference at k=3 is 100 minus 2 times 70 plus 60, which equals 20. At k=4 it is 70 minus 2 times 60 plus 58, which equals 8. Under the stated maximum-second-difference convention, k=3 marks the stronger elbow.

A direct-minimum convention applied to the same finite candidate list would choose a different k, showing why software documentation is part of the mathematics. Determinants also depend on variable scaling and can collapse under singularity. Report T and W definitions, determinant method, k range, hierarchy, penalty, difference convention and numerical safeguards.

---

Frequently asked questions

What is the Marriott index?

A determinant-based clustering criterion commonly written as k squared times the pooled within-cluster determinant.

Why multiply by k squared?

It counters the routine determinant reduction that occurs when more clusters are permitted.

Which direction is better?

The historical raw criterion and software curvature rules must be stated separately; this article demonstrates a maximum-second-difference convention.

Why can conventions disagree?

Direct minima and discrete curvature answer different questions and can select different k values on the same range.

What if W is singular?

The determinant is zero and can dominate the criterion; regularisation changes the calculation.

What must be reported?

W definition, determinant method, k range, penalty, raw sequence, first and second differences, selection rule and sensitivity checks.

---

Primary source and fact-check note

F. H. C. Marriott, “Practical Problems in a Method of Cluster Analysis,” Biometrics 27(3), 501–514 (1971), DOI 10.2307/2528592. Definitions and core claims in this article were checked against this primary publication. Calculations shown here were independently recomputed. Examples are educational illustrations; they are not reported industrial performance results.

Useful next reading

• Scott–Symons log-determinant ratio
• Rubin determinant ratio
• SD validity range dependence
• How AI finds patterns without labels
• eduKateSingapore learning routes

Keep the options open. Mathematics education supports clearer questions, transferable problem-solving skills and informed choices across science, engineering, computing, AI, statistics, design and many other careers. It does not guarantee intelligence, income, admission or employment; those outcomes depend on far more than one subject or model.

Extended laboratory: matrix evidence and responsible cluster selection

Data audit

Freeze observations, variables, units, exclusions, missing-data rules, weights, standardisation, sample size and seeds.

Scatter audit

Define total, within and between sums-of-squares and cross-products matrices with centring and denominators.

Linear-algebra audit

Record determinants, log determinants, eigenvalues, condition numbers, inverse method and any numerical tolerance.

Sequence audit

Show raw scores, first differences, centred second differences, candidate labels and selection direction.

Convention audit

Separate historical optimisation rules from current software conventions and name the implementation and version.

Uncertainty audit

Repeat starts, samples, variable sets, transformations, candidate ranges and plausible regularisation settings.

Meaning audit

Compare internal geometry with stability and domain evidence without assuming clusters are natural categories.

Final claim

State the exact matrix evidence and stop before causal, semantic, career or guaranteed-performance claims.

Mechanism workshop: calculate, challenge and transfer

Workshop 1: Criterion

Begin with this mechanism: Define Mk=k^2 det(Wk). Recreate the concrete case: Penalty and generalised dispersion are explicit. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Spell Marriott consistently even if software says marriot. Complete the transfer step: State formula. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 2: Two clusters

Begin with this mechanism: Set det(W2)=25. Recreate the concrete case: M2=4×25=100. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Use squared k, not k. Complete the transfer step: Show step. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 3: Three clusters

Begin with this mechanism: Set det(W3)=70/9. Recreate the concrete case: M3=9×70/9=70. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Retain exact fractions where useful. Complete the transfer step: Show step. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 4: Four clusters

Begin with this mechanism: Set det(W4)=3.75. Recreate the concrete case: M4=16×3.75=60. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Same scatter convention is required. Complete the transfer step: Show step. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 5: Five clusters

Begin with this mechanism: Set det(W5)=2.32. Recreate the concrete case: M5=25×2.32=58. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: A smaller determinant is expected with larger k. Complete the transfer step: Show step. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 6: Raw sequence

Begin with this mechanism: List 100,70,60,58. Recreate the concrete case: Decline slows after k=3. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Do not choose terminal k automatically. Complete the transfer step: Plot sequence. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 7: First difference one

Begin with this mechanism: Compute 70-100. Recreate the concrete case: Delta3=-30. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Index differences need an order convention. Complete the transfer step: Record value. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 8: First difference two

Begin with this mechanism: Compute 60-70. Recreate the concrete case: Delta4=-10. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Negative means the criterion fell. Complete the transfer step: Record value. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 9: First difference three

Begin with this mechanism: Compute 58-60. Recreate the concrete case: Delta5=-2. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: The sequence is flattening. Complete the transfer step: Record value. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 10: Second difference at three

Begin with this mechanism: Compute 100-2×70+60. Recreate the concrete case: D2 at k=3 is 20. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Centre the label correctly. Complete the transfer step: Show step. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 11: Second difference at four

Begin with this mechanism: Compute 70-2×60+58. Recreate the concrete case: D2 at k=4 is 8. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Use identical rounding. Complete the transfer step: Show step. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 12: Software selection

Begin with this mechanism: Compare 20 and 8. Recreate the concrete case: Maximum second difference marks k=3. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: This is a declared convention. Complete the transfer step: State result. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 13: Historical minimum

Begin with this mechanism: Inspect k^2 det(Wk) directly. Recreate the concrete case: A direct-minimum rule can disagree on a finite range. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Do not merge conventions. Complete the transfer step: Report both. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 14: Singularity

Begin with this mechanism: Let W lose rank. Recreate the concrete case: det(W)=0 can dominate the score. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Regularisation changes the criterion. Complete the transfer step: Flag failure. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 15: Candidate range

Begin with this mechanism: Extend k beyond five. Recreate the concrete case: New curvature or minima may appear. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Range is part of the evidence. Complete the transfer step: Run sensitivity. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 16: Claim

Begin with this mechanism: Name W,k range,penalty and difference rule. Recreate the concrete case: Teaching curvature supports k=3. Write k, Wk, determinant, k-squared penalty, raw score, first differences, second differences and chosen k with identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Do not promise natural groups. Complete the transfer step: Bound conclusion. Change exactly one sample size, variable, scale, cluster count, scatter entry, determinant, eigenvalue, matrix condition, regularisation constant, candidate range, first difference, second difference, initialisation or seed. Explain which mathematical object changes, which score moves and which interpretation must weaken. Keep fit, penalty, curvature, conditioning, stability, uncertainty and practical usefulness on separate notebook lines. Explain once to a student and once to a parent. Add a counterexample involving a constant variable, collinearity, singular W, tiny eigenvalue, singleton cluster or boundary candidate. State the extra evidence needed to distinguish explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.