DataTerminology.comNEW
A functional relationship where one quantity varies as a constant times another quantity raised to a power.
DESCRIPTION
Power laws describe phenomena where relative changes scale proportionally, producing heavy tails and scale invariance. Examples appear in city sizes, word frequencies, and network degrees. On log-log plots, power laws appear as straight lines, simplifying parameter estimation. Distinguishing power laws from log-normal or other heavy-tailed distributions requires rigorous statistical tests. Parameter estimation often uses maximum likelihood with goodness-of-fit comparisons. Misidentifying power-law behavior is common without careful modeling. Understanding tail behavior informs risk management, resilience, and resource allocation.
ALTERNATIVE DEFINITIONS
Wikipedia
A power law is a functional relationship between two quantities, where a relative change in one quantity results in a proportional relative change in the other quantity, independent of the initial size of those quantities.
Highlights proportional relative changes and scale invariance.
Wolfram MathWorld
A power law is a relationship of the form y = C x^a, where C and a are constants.
Emphasizes the canonical algebraic form of power laws.
RELATED TERMS
Heavy-Tailed Distribution Pareto Distribution Zipf’s Law Scale-Free Network Log-Log Plot Cumulative Distribution Function Tail Exponent
COMPETING TERMS
Normal Distribution Exponential Distribution
KNOWLEDGE AREAS
TARGET AUDIENCE
Data Scientist Data Analyst Information Analyst Chief Data Officer
EXTERNAL SOURCES
- Wikipedia - Power law
- SIAM Review - Power-law distributions in empirical data
- Santa Fe Institute - Resources on power-law analysis
TOP QUESTIONS
Why are Power Laws significant in data science?
They model heavy-tailed behavior and inequality in real systems, informing risk, resource allocation, and network robustness.
Category: Relevancy
How do I identify a Power Law?
Plot data on log-log axes and test with maximum likelihood fits and goodness-of-fit statistics against alternatives like log-normal.
Category: HowTo
What’s a common Power Law misconception?
Straight lines on log-log plots can arise from non-power-law distributions. Always validate with statistical tests.
Category: Misunderstanding
Where do Power Laws appear in practice?
City sizes, word frequencies (Zipf’s law), wealth distributions, and network degrees in scale-free graphs.
Category: Example
What is the tail exponent in a Power Law?
It’s the parameter controlling tail heaviness; smaller exponents indicate heavier tails and higher variance.
Category: Explanation
How do Power Laws impact risk management?
Heavy tails mean rare, extreme events are more probable, requiring plans for outliers and resilience.
Category: Risks
Are Power Laws scale-invariant?
Yes. Rescaling variables preserves the relationship form, a hallmark of power-law behavior.
Category: Explanation
What tools help fit Power Laws?
Use MLE methods, Clauset-Newman-Shalizi procedures, and statistical packages implementing tail-fitting diagnostics.
Category: HowTo
What is Zipf’s Law in relation to Power Laws?
Zipf’s law is a specific power-law relation describing rank-frequency distributions, notably in language.
Category: Explanation