Two tracks · anatomy and mathematics

Learn

Two tracks, one goal: that you leave able to build this yourself. Inside the pipeline pulls apart the recollection-interval workflow behind the catalog, one component at a time, each with working R and Python implementations to assemble into your own script. Concepts explains the mathematics underneath — why each step is shaped the way it is. Take either track in order, or jump between them through the cross-links on every page.

Track 1

Inside the pipeline

The workflow that turns long-term germination monitoring into recollection intervals, storage recommendations, and freeze-sensitivity flags — decomposed into twelve components. Each page: what the component does, why it's designed that way, and code to make it yours.

  1. 01Read messy exports, map columns by roleR · Python
  2. 02Turn two dates into one clockR · Python
  3. 03Counts are the truth; percentages are a courtesyR · Python
  4. 04Many tests, one year: max or poolR · Python
  5. 05The early-window baseline: what 70% is 70% ofR · Python
  6. 06The observed interval: no model, just the latest year that cleared the barR · Python
  7. 07Fitting the decline: probit regression that survives perfect dataR · Python
  8. 08Observed vs supported: two independent confidence checksR · Python
  9. 09From accessions to a species numberR · Python
  10. 10R5 vs R18: detecting seeds that freezing hurtsR · Python
  11. 11Potential supporters: where to look nextR · Python
  12. 12Censoring-honest notation and the publication sheetR · Python
Track 2

Concepts

The twelve mathematical ideas the pipeline (and the rest of this lab) stands on, one per page, in dependency order — from a single germination test as binomial evidence up to comparing decay models.

  1. 01A germination test is binomial evidenceinteractive
  2. 02The clock: an accession's age at test
  3. 03Replicates: same seeds, same year, different answers
  4. 04The baseline questioninteractive
  5. 05Three different P70sinteractive
  6. 06Observed intervals, censoring, and honest notation
  7. 07Fitting the decline: probits, σ, and Ki
  8. 08Support: when the uncertainty agrees with the point
  9. 09Accessions to species: the maximum, qualified
  10. 10Comparing storages and catching freeze damage
  11. 11From numbers to recommendations
  12. 12Comparing models: AICc and the discipline of parsimony
Reference shelf
Method — the lab's own fitting choices (σ-primary, validation against the field-standard R implementation, the baseline question) · Sources — every federated dataset with its license and what we do and don't redistribute.