Biology Atlas

Map / Open register

question · intermediate · partly written

How noisy is a reaction yield between repeats, and does adding failed reactions help prediction models?

No numeric yield noise between repeats was found, and whether failed reactions help yield models is not settled.

What is measured about yield and conditions?

Yield prediction had mean absolute error 15.6% against 16.3% for guessing the mean, and over half of 15.4 million filtered Reaxys reactions had no yield recorded. 12

Reported, not yet replicated

Why is the noise figure missing? more

Inter-lab variation is described only in words in the sources opened, and failed reactions are rarely published, so a clean experiment needs new data. 1

Our inference

Observation

Yield models barely beat guessing the mean (mean absolute error 15.6% against 16.3%), and failed reactions are probably near absent from public sets. 13

Reported, not yet replicated

What is known

Literature and patent yields are noisy and condition data are missing for most Reaxys records (54% no yield, 53.9% no temperature). 21

Reported, not yet replicated

What is not

A measured yield variation between repeats or labs, and a controlled test of adding failures, were not found in the sources opened. 1

Our inference

What would settle it

Publish repeated runs of one reaction under stated conditions, then train yield models with and without recorded failures against the guess-the-mean baseline. This is our own suggestion. 1

Our inference

Still open here

Not yet written: Hypotheses, What is missing, Who works on it.

Sources

  1. Beker et al.. Machine learning may sometimes simply capture literature popularity trends [title not copied from the page]. journal not copied (2022). paper; opened 2026-11-17 all records citing this
  2. Jacob et al.. Analysis of Reaxys reaction records for missing conditions [title not copied from the page]. Journal of Cheminformatics (2017). paper; opened 2026-11-17 all records citing this
  3. Connor W. Coley, et al.. Computer-assisted retrosynthesis based on molecular similarity. ACS Central Science (2017). paper; opened 2026-11-17 all records citing this
Where it sits on the map
History of this record (1)

This record has been changed 1 time, on the dates below. The note kept against each change is written for the people who maintain the atlas, in their own shorthand, so it is not shown here; it is in the record itself.

Kind
big-open
Register
open-question
Triage
needs-research
Status
partly written
Last updated
2026-11-17
Record
content/questions/how-noisy-is-a-repeated-reaction-yield-and-do-failed-reactions-help.json