Choose your reading route
eduKateSG · Why Mathematics? · Local Block Bootstrap: Slowly Changing Time-Series Structure
Start with the question, audit the worked example, or jump to the assumptions and practice. Every chapter stays open for uninterrupted reading.
Understand the question
Define the quantity and follow the mechanism.Route 2 · Chapters 8–14
Reproduce the example
Audit the numbers one step at a time.Route 3 · Chapters 15–21
Check assumptions
See where design and calibration matter.Route 4 · Chapters 22–28
Explore applications
Connect the method to a concrete investigation.Route 5 · Chapters 29–34
Study and verify
Practise, inspect failure cases and read further.
Full chapter index · Practice and checking · Mathematics Learning Hub
Why is mathematics important when a time series changes slowly, so blocks from distant dates are not exchangeable? The local block bootstrap resamples blocks from neighbourhoods near each target time. It preserves short-range dependence while allowing the data-generating structure to evolve across the record.
Did You Know? The method needs two scales: block length for dependence and locality width for gradual change. A good block can still be drawn from the wrong era if the neighbourhood is too wide.
Full chapter index
Chapters 1–7 · Understand the question
Chapters 8–14 · Reproduce the example
Chapters 15–21 · Check assumptions
Chapters 22–28 · Explore applications
Chapters 29–34 · Study and verify
Global block bootstrap assumes stationarity: a segment from the beginning can replace one near the end. Trends, evolving variance or gradual seasonal changes violate that exchangeability.
Paparoditis and Politis defined local block resampling for stochastic structure that changes slowly with time. Read the article abstract.
For a target position t, choose candidate block starts whose rescaled times lie within a locality bandwidth around t. Draw among those candidates rather than across the full record.
With n=100 and locality half-width 10, target t=60 uses candidate starts approximately 50 through 70, subject to block and boundary rules.
| n | Target | Half-width | Starts |
|---|---|---|---|
| 100 | 60 | 10 | 50–70 |
If block length l=5 and start 64 is selected, the copied block is positions 64–68. The next target location then chooses from its own local candidate set.
| Start | l | Positions |
|---|---|---|
| 64 | 5 | 64–68 |
Block length controls preserved dependence; locality bandwidth controls how quickly the resampling law may change. Their errors interact and should be tuned jointly.
Local stationarity means a short neighbourhood behaves approximately stationary while its parameters evolve over longer time. Sudden breaks need segmentation or break-aware methods.
Means or trends may be estimated locally before resampling. Using a global centre can reintroduce the nonstationarity the method was designed to respect.
Near time 1 or n, symmetric neighbourhoods are impossible. One-sided, reflected or adaptive windows change candidate probabilities and must be reported.
Locality should be reported in observations and real time. A width of 20 means twenty minutes for one sensor and twenty months for another.
A local window must include appropriate seasonal phases or use seasonal alignment. Nearby calendar times can still be incomparable across weekday or season.
Local resampling can preserve gradual variance changes if neighbourhoods are narrow enough. Too-wide windows mix low- and high-volatility eras.
Estimate uncertainty of a smooth time-varying mean by generating local block paths and recalculating the mean curve. Simultaneous bands need joint calibration.
The most recent neighbourhood may inform a near-future forecast better than old segments, but extrapolation beyond the observed evolution remains a modelling assumption.
Gradually changing seasonal amplitude may suit local blocks after careful detrending. Long-term forced trends should not be interpreted as resampling noise.
Moving blocks sample candidate segments globally under stationarity. Local blocks restrict candidates by target time to accommodate slow evolution.
A rolling analysis fits separate estimates to recent data. Local block bootstrap generates whole pseudo-series using time-varying candidate neighbourhoods.
Local methods assume gradual evolution. Change-point methods estimate abrupt regimes; mixing the two without a model can blur genuine breaks.
Plot candidate windows, local means, variances and autocorrelations. Check how often each original time is selected and whether generated paths preserve evolution.
Simulate locally stationary processes and abrupt-break alternatives, vary both bandwidths, measure coverage and compare with global blocks and segmented bootstraps.
Local blocks can fail when change is abrupt, neighbourhoods are sparse, seasonality is misaligned, dependence exceeds l or future evolution is extrapolated without support.
Create rows for target time, local candidate range, selected start, copied block and boundary adjustment. Plot candidate coverage across the record.
Unit-test n=100, t=60, width 10 gives candidates 50–70 and l=5 at start 64 gives 64–68. Save both tuning parameters and seed.
Local resampling matters in climate, finance, health monitoring and sensors. Careers require nonstationary time-series reasoning, computation and scientific context.
Simulate a gradually rising AR variance and an abrupt break. Compare global and local block coverage across locality and block grids.
Local does not mean independent, nearby times are not always seasonally comparable, and gradual-change methods do not automatically detect breaks.
CHAPTER 27 OF 34 · Explore applications
27. Connection to recent mathematics articles
The moving-block guide explains global blocks. The block-length guide explains dependence tuning.
Students should draw candidate windows on a timeline. Parents can ask which changes are treated as gradual structure and which are treated as random variation.
- Find candidates 50–70.
- List block 64–68.
- Separate both bandwidths.
- Diagnose an abrupt break.
- Handle boundaries.
### Is the full record exchangeable? No.
### How many tuning scales? At least block and locality widths.
### Does it model abrupt breaks? Not by itself.
### Why use real time units? Sampling intervals differ.
Report time scale, local-stationarity assumption, block length, locality width, candidate rule, edges, seasonality, B, seed, diagnostics and break sensitivity.
The dependence block determines what moves together; the locality window determines where that block may come from. Confusing these roles produces plausible-looking but invalid paths.
Confirm slow evolution, separate tuning scales, valid seasonal matching, transparent edges, candidate-coverage diagnostics and explicit failure under abrupt change.
Continue through the Mathematics Learning Hub. Rebuild one block, taper, local-neighbour or regeneration calculation by hand before trusting automated output.
Extended practice laboratory
Practice case 1: Plot evolving series
Setup: Start from the article’s local-block mechanism for case 1. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 2: Choose block length
Setup: Start from the article’s local-block mechanism for case 2. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 3: Choose locality width
Setup: Start from the article’s local-block mechanism for case 3. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 4: Select target time
Setup: Start from the article’s local-block mechanism for case 4. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 5: Find candidate starts
Setup: Start from the article’s local-block mechanism for case 5. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 6: Draw start
Setup: Start from the article’s local-block mechanism for case 6. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 7: Copy block
Setup: Start from the article’s local-block mechanism for case 7. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 8: Advance target
Setup: Start from the article’s local-block mechanism for case 8. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 9: Handle left edge
Setup: Start from the article’s local-block mechanism for case 9. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 10: Handle right edge
Setup: Start from the article’s local-block mechanism for case 10. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 11: Use local centring
Setup: Start from the article’s local-block mechanism for case 11. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 12: Align season
Setup: Start from the article’s local-block mechanism for case 12. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 13: Model trend
Setup: Start from the article’s local-block mechanism for case 13. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 14: Model variance
Setup: Start from the article’s local-block mechanism for case 14. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 15: Increase locality
Setup: Start from the article’s local-block mechanism for case 15. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 16: Reduce locality
Setup: Start from the article’s local-block mechanism for case 16. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 17: Increase block
Setup: Start from the article’s local-block mechanism for case 17. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 18: Add gradual change
Setup: Start from the article’s local-block mechanism for case 18. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 19: Add abrupt break
Setup: Start from the article’s local-block mechanism for case 19. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 20: Compare global blocks
Setup: Start from the article’s local-block mechanism for case 20. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 21: Change seed
Setup: Start from the article’s local-block mechanism for case 21. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 22: Increase B
Setup: Start from the article’s local-block mechanism for case 22. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 23: Audit candidates
Setup: Start from the article’s local-block mechanism for case 23. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 24: Name limits
Setup: Start from the article’s local-block mechanism for case 24. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.
Practice case 25: Write conclusion
Setup: Start from the article’s local-block mechanism for case 25. Task: Change the named quantity, recompute the dependent result and compare with the original.
Check: Preserve the dependence unit, verify the weight, window or cycle identity, and state one diagnostic before interpretation. Record the units and compare with one defensible alternative. Show the arithmetic, name the response and predictor units, and separate the numerical result from the claim it supports. Then vary one defensible assumption, calculate again and explain whether the conclusion changed because of the data, the fitted model or the validation purpose. Finish with one compact table or graph that could reveal a failure hidden by the summary.