WHY MATHEMATICS?

Why Mathematics? | Functional Linear Quantile Regression: Curve Predictors, Principal Components, Inner Products and Tail-Specific Effects

How a whole sensor or growth curve predicts an outcome percentile through a tau-specific slope function, and why basis choice, noisy grids and crossing require care.

Why mathematics is important becomes easier to see when a real decision depends on it. This guide follows the mechanism from definitions and dimensions to a worked calculation, validation, practical learning and responsible use. It keeps factual evidence separate from interpretation and treats every illustrative scenario as an example rather than a promise of results.

Choose a reading route

Represent each subject by a centred functional predictor
Project curves into a functional principal-component basis
Fit check loss with a tuned component cutoff
Validate reconstruction, quantile order and tail-specific slope stability

Chapter map

1. Functional predictor
2. Domain
3. Centering
4. Target
5. Slope function
6. Quadrature
7. First term
8. Middle term
9. Last term
10. Inner product
11. Intercept
12. Functional PCA
13. Cutoff
14. Measurement error
15. Quantile order
16. Claim

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

1. Functional predictor

Here the geometry becomes operational. Treat one curve as one observation.

A concrete situation

Use values 1,2,1 over time.

What the mathematics does not license

Columns are not independent cases. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Keep subject IDs. 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. Domain

This step deserves a careful pause. Set times 0,.5,1.

A concrete situation

Use a common interval.

What the mathematics does not license

Irregular grids need reconstruction. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Record times. 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. Centering

The useful question is not only how to calculate, but what the calculation assumes. Subtract the training mean function.

A concrete situation

Teaching curve is already centred.

What the mathematics does not license

Using full-data mean leaks validation. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Fit within folds. 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. Target

A small example makes the mechanism visible. Choose tau and scalar outcome.

A concrete situation

Use one conditional quantile.

What the mathematics does not license

Slope function can vary with tau. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Fit a grid later. 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.

Did You Know? In this chapter, the most important number is useful only together with its construction rule. Changing scaling, sample membership or validation design can change the numerical answer without changing the arithmetic.

5. Slope function

This is where modelling choices begin to affect evidence. Use values .5,-.5,1.

A concrete situation

Middle time has negative association.

What the mathematics does not license

Pointwise signs can be unstable. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Use simultaneous view. 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. Quadrature

The central mathematical idea is concrete. Use weights .25,.5,.25.

A concrete situation

Weights sum to one interval length.

What the mathematics does not license

Wrong grid weights distort integral. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Check geometry. 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 term

Here the geometry becomes operational. Compute .25 times 1 times .5.

A concrete situation

Get .125.

What the mathematics does not license

Keep sign and units. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Show work. 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. Middle term

This step deserves a careful pause. Compute .5 times 2 times -.5.

A concrete situation

Get -.5.

What the mathematics does not license

Largest contribution is negative. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Preserve sign. 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. Last term

The useful question is not only how to calculate, but what the calculation assumes. Compute .25 times 1 times 1.

A concrete situation

Get .25.

What the mathematics does not license

Endpoint weight differs from midpoint. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Use trapezoid. 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. Inner product

A small example makes the mechanism visible. Add the three terms.

A concrete situation

Get -.125.

What the mathematics does not license

Discrete approximation is not exact integral. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Increase grid if justified. 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. Intercept

This is where modelling choices begin to affect evidence. Add beta_0=3.

A concrete situation

Fitted conditional quantile is 2.875.

What the mathematics does not license

One curve does not estimate a model. The same reasoning helps students move from an exercise to a real dataset.

Try it, check it, transfer it

Label illustration. 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. Functional PCA

The central mathematical idea is concrete. Project curves onto eigenfunctions.

A concrete situation

Retain K scores.

What the mathematics does not license

Components with high variance need not predict the target tail. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Tune K. 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. Cutoff

Here the geometry becomes operational. Balance truncation bias and variance.

A concrete situation

Use cross-validation inside training data.

What the mathematics does not license

Too many components amplify noise. The habit of checking assumptions is one of the practical benefits of learning mathematics.

Try it, check it, transfer it

Plot stability. 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. Measurement error

This step deserves a careful pause. Smooth noisy sparse curves.

A concrete situation

Propagate reconstruction uncertainty.

What the mathematics does not license

Treating estimates as exact understates error. The same reasoning helps students move from an exercise to a real dataset.

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.

15. Quantile order

The useful question is not only how to calculate, but what the calculation assumes. Check fitted curves across tau.

A concrete situation

Monotonise when necessary.

What the mathematics does not license

Repair changes estimates and uncertainty. That distinction matters because a correct formula can still answer the wrong operational question.

Try it, check it, transfer it

Report method. 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

A small example makes the mechanism visible. State domain, basis, K, tau and preprocessing.

A concrete situation

Teaching prediction is 2.875.

What the mathematics does not license

Do not call slope curve causal timing. The habit of checking assumptions is one of the practical benefits of learning mathematics.

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

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

Functional linear quantile regression uses an entire curve as a predictor for a scalar outcome quantile. A daily sensor trace, growth curve or repeated signal is treated as one functional covariate rather than as dozens of unrelated columns. Kato studied a model in which the tau conditional quantile equals an intercept plus the inner product of a centred curve and a tau-specific slope function, estimated through a functional principal-component basis.

Suppose a centred curve is observed at times 0,.5,1 with values 1,2,1, and a tau-specific slope function has values .5,-.5,1. Using trapezoidal weights .25,.5,.25, the approximate inner product is .25(1)(.5)+.5(2)(-.5)+.25(1)(1)=-.125. With intercept 3, the fitted conditional quantile is 2.875. The sign and timing of the slope function determine which parts of the curve raise or lower the target quantile.

The functional principal components are estimated from data, then truncated. Too few components omit predictive directions; too many amplify noise and make slope curves unstable. Irregular sampling, measurement error, centring and sparse curves require explicit reconstruction choices. When several tau levels are fitted separately, plug-in quantile curves can cross and may need a documented monotonisation step.

Report the curve domain, measurement grid, smoothing, centring, functional basis, component cutoff, tau grid, check-loss fitting, crossing repair, validation folds and uncertainty. Estimate preprocessing only inside each training fold. A colourful slope curve is not a causal time map, and unstable regions with little functional variation should not be over-interpreted.

Frequently asked questions

What is a functional predictor?

A curve or signal treated as one covariate object for each subject.

How does the curve enter the model?

Through an inner product with a tau-specific slope function.

Why use functional principal components?

They reduce an infinite-dimensional curve to a finite set of scores.

How is the component cutoff chosen?

With training-only validation that balances truncation bias and noise.

Can fitted quantiles cross?

Yes. Separate plug-in fits may need a documented monotonisation step.

What must be reported?

Curve domain, smoothing, centring, basis, cutoff, tau grid, repair, uncertainty and support.

Primary source and fact-check note

Kengo Kato, “Estimation in Functional Linear Quantile Regression,” The Annals of Statistics 40(6), 3108–3136 (2012), DOI 10.1214/12-AOS1066. 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

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: topic-specific validation and responsible use

Design audit

Freeze population, sample unit, outcome scale, curve identity, response threshold grid, quantile convention, exposure, weights, missingness and dependence.

Transformation audit

Write the RIF, binary threshold indicators, jitter mapping or functional inner product explicitly and preserve original units.

Tuning audit

Record density bandwidth, response thresholds, probability link, jitter replications, curve smoother, basis and component cutoff.

Computation audit

Check indicator equality, CDF order, random seeds, integer mapping, quadrature weights, PCA leakage and solver tolerance.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Coherence audit

Inspect marginal versus conditional targets, monotonicity, discrete jumps, curve support, quantile crossing and subgroup calibration.

Uncertainty audit

Propagate density, binary-fit, jitter, curve reconstruction, tuning and sampling uncertainty through the complete pipeline.

Validation audit

Use untouched observations or subjects for probability calibration, check loss, count coverage, centile order and sensitivity.

Final claim

Translate the distributional result without turning a projection, auxiliary variable, fitted curve or predictive association into a guarantee.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Mechanism workshop: calculate, challenge and transfer

Workshop 1: Functional predictor

Begin with this mechanism: Treat one curve as one observation. Recreate the concrete case: Use values 1,2,1 over time. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Columns are not independent cases. Complete the transfer step: Keep subject IDs. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 2: Domain

Begin with this mechanism: Set times 0,.5,1. Recreate the concrete case: Use a common interval. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Irregular grids need reconstruction. Complete the transfer step: Record times. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 3: Centering

Begin with this mechanism: Subtract the training mean function. Recreate the concrete case: Teaching curve is already centred. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Using full-data mean leaks validation. Complete the transfer step: Fit within folds. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 4: Target

Begin with this mechanism: Choose tau and scalar outcome. Recreate the concrete case: Use one conditional quantile. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Slope function can vary with tau. Complete the transfer step: Fit a grid later. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Workshop 5: Slope function

Begin with this mechanism: Use values .5,-.5,1. Recreate the concrete case: Middle time has negative association. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Pointwise signs can be unstable. Complete the transfer step: Use simultaneous view. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 6: Quadrature

Begin with this mechanism: Use weights .25,.5,.25. Recreate the concrete case: Weights sum to one interval length. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Wrong grid weights distort integral. Complete the transfer step: Check geometry. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 7: First term

Begin with this mechanism: Compute .25 times 1 times .5. Recreate the concrete case: Get .125. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Keep sign and units. Complete the transfer step: Show work. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 8: Middle term

Begin with this mechanism: Compute .5 times 2 times -.5. Recreate the concrete case: Get -.5. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Largest contribution is negative. Complete the transfer step: Preserve sign. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Workshop 9: Last term

Begin with this mechanism: Compute .25 times 1 times 1. Recreate the concrete case: Get .25. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Endpoint weight differs from midpoint. Complete the transfer step: Use trapezoid. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 10: Inner product

Begin with this mechanism: Add the three terms. Recreate the concrete case: Get -.125. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Discrete approximation is not exact integral. Complete the transfer step: Increase grid if justified. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 11: Intercept

Begin with this mechanism: Add beta_0=3. Recreate the concrete case: Fitted conditional quantile is 2.875. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: One curve does not estimate a model. Complete the transfer step: Label illustration. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 12: Functional PCA

Begin with this mechanism: Project curves onto eigenfunctions. Recreate the concrete case: Retain K scores. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Components with high variance need not predict the target tail. Complete the transfer step: Tune K. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Workshop 13: Cutoff

Begin with this mechanism: Balance truncation bias and variance. Recreate the concrete case: Use cross-validation inside training data. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Too many components amplify noise. Complete the transfer step: Plot stability. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 14: Measurement error

Begin with this mechanism: Smooth noisy sparse curves. Recreate the concrete case: Propagate reconstruction uncertainty. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Treating estimates as exact understates error. Complete the transfer step: Run sensitivity. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 15: Quantile order

Begin with this mechanism: Check fitted curves across tau. Recreate the concrete case: Monotonise when necessary. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Repair changes estimates and uncertainty. Complete the transfer step: Report method. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.

Workshop 16: Claim

Begin with this mechanism: State domain, basis, K, tau and preprocessing. Recreate the concrete case: Teaching prediction is 2.875. Write subject IDs, curve times, curve values, centring functions, basis scores, eigenfunctions, slope functions, quadrature weights, inner products, quantile levels and scalar outcomes with units, signs, identifiers and assumptions. Calculate the smallest non-trivial example by hand and reproduce it independently. Challenge this limitation: Do not call slope curve causal timing. Complete the transfer step: Bound conclusion. Change exactly one outcome, quantile convention, density bandwidth, indicator equality rule, population weight, threshold, binary link, monotonicity repair, jitter draw, random seed, exposure, integer mapping, curve time, smoothing method, basis, component cutoff, quadrature weight, tau or validation sample. Explain which mathematical object changes, which estimator is selected, which conclusion survives and which claim must weaken. Keep design, transformation, tuning, computation, coherence, uncertainty and interpretation on separate notebook lines. Explain the result once to a student with the complete table and once to a parent with the real decision and original units. Add a counterexample involving sparse density, threshold separation, a decreasing CDF, zero inflation, irregular curves, measurement error, subgroup miscalibration or support failure. State what extra measurement would separate the explanations. Finish with one supported sentence, one unsupported sentence and one useful next measurement.