Editing Learning curve (section)

=== Models ===
[[File:Learning curve models- Wright, Plateau, Stanford-B, DeJong, S-curve.svg|thumb|307x307px|The main learning curve models on a log-log plot. Wright, Plateau, Stanford-B, DeJong, S-curve.]]
The main statistical models for learning curves are as follows:<ref name=":0">{{Cite journal |last=Yelle |first=Louis E. |title=The Learning Curve: Historical Review and Comprehensive Survey |date=April 1979 |url=https://onlinelibrary.wiley.com/doi/10.1111/j.1540-5915.1979.tb00026.x |journal=Decision Sciences |language=en |volume=10 |issue=2 |pages=302–328 |doi=10.1111/j.1540-5915.1979.tb00026.x |issn=0011-7315|url-access=subscription }}</ref><ref>{{Cite journal |last1=Anzanello |first1=Michel Jose |last2=Fogliatto |first2=Flavio Sanson |date=2011-09-01 |title=Learning curve models and applications: Literature review and research directions |url=https://www.sciencedirect.com/science/article/pii/S016981411100062X |journal=International Journal of Industrial Ergonomics |language=en |volume=41 |issue=5 |pages=573–583 |doi=10.1016/j.ergon.2011.05.001 |issn=0169-8141|url-access=subscription }}</ref>

* Wright's model ("log-linear"): <math>y = Kx^n</math>, where 
** <math>y</math> is the cost of the <math>x</math>-th unit, 
** <math>x</math> is the total number of units made,
** <math>K</math> is the cost of the first unit made,
** <math>n</math> is the exponent measuring the strength of learning. 
* Plateau model: <math>y = \max(Kx^n, K_0)</math>, where <math>K_0</math> models the minimal cost achievable. In other words, the learning ceases after cost reaches a sufficiently low level.
* Stanford-B model: <math>y = K(x+B)^n</math>, where <math>B</math> models worker's prior experience.
* DeJong's model: <math>y = K(M + (1-M)x^n)</math>, where <math>M</math> models the fraction of production done by machines (assumed to be unable to learn, unlike a human worker).
* S-curve model: <math>y = K(M + (1-M)(x+B)^n)</math>, a combination of Stanford-B model and DeJong's model.

The key variable is the exponent <math>n</math> measuring the strength of learning. It is usually expressed as <math>n = \log(\phi)/\log(2)</math>, where <math>\phi</math> is the "learning rate". In words, it means that the unit cost decreases by <math>1-\phi</math>, for every doubling of total units made. Wright found that <math>\phi \approx 80\%</math> in aircraft manufacturing, meaning that the unit cost decreases by 20% for every doubling of total units made.