“Rarefaction Engine”


  

Tutorial


The “rarefaction engine“ measures diversity

The “rarefaction engine“ measures the mathematical diversity of an array of integers (often used to designate other data types, e.g. 1=“apple“, 2=“lemon“). It outputs the mean number of unique integers found within the array itself when subjected to repeated, random sampling of a fixed size. The sample size is then gradually increased in a stepwise fashion, resulting in a series of means known as the “rarefaction line“. This line can then be drawn on a graph and compared to lines produced by other datasets.

The “rarefaction engine“ excels where arithmetics fail

The “rarefaction engine“ excels when the diversity within two distributions are compared in situations where there is no arithmetically meaningful way to do so, e.g. when comparing vast numbers of categories, or when the size of the underlying population is simply too great to measure. Sadly, the downside of this type of within-group analysis is that it cannot be used to generate a meaningful statistical inference.

Extensible columns, extensible rows

The “rarefaction engine“ offers extensibility in both columns and rows to suit your needs when analyzing a table of data.

1. Demo tables

Demo type Range (rows : columns)
Binomial
Binomial (6 singletons lost)
Binomial (6 singletons relocated)
Even
Symmetrical (columns enumerated)
Symmetrical (rows enumerated)
Homogeneous

2. Monitor

3. Define probe

Top-most proportion of rarefaction width covered:              

Probe 1 2 3 4 5 6 7 8 9 10 11 12
Sample size

4. Rarefaction

Number of rarefaction iterations:           

Probe 1 2 3 4 5 6 7 8 9 10 11 12
Rarefaction

5. Diagnostics

Contact us



Magnus Olofsson, MD
Ödeshög Health Care Centre
Skolgatan 3-7
SE-599 31  ÖDESHÖG
SWEDEN
Telephone: (+46) 10-104 87 16
E-mail: magnus.olofsson@lio.se