### Introduction

### Theory, R functions & Examples

en:similarity_r

- offers just limited number of distance measures - e.g.`dist`

`euclidean`

,`canberra`

and`manhattan`

. The result is the*distance matrix*, an object of the class`dist`

.- default distance used in this function is Bray-Curtis distance, which is (in contrast to Euclidean distance) considered as more suitable for ecological data (it is a quantitative analog of Sørensen dissimilarity).`vegdist`

(library`vegan`

)- offers some other indices than`dsvdis`

(library`labdsv`

)`vegdist`

, e.g.`ruzicka`

(Růžička, quantitative analogue of Jaccard) and`roberts`

. For full comparison of`dist`

,`vegdist`

and`dsvdis`

, see Table 1 in the website of Dave Roberts.- includes 21 dissimilarity indiced described in Legendre & De Cáceres (2013), twelve of which are not readily available in other packages. Note that Bray-Curtis dissimilarity is called`dist.ldc`

(library`adespatial`

)*percentage difference*(`method = “percentdiff”`

). By default returns also informative message whether given dissimilarity index is Euclidean or not and whether it becomes Euclidean if square-rooted (as is the case of e.g. Bray-Curtis aka Percentage difference index).- allows to design virtually any distance measure using the formula for their calculation.`designdist`

(library`vegan`

)- offers euclidean, manhattan and gower distance.`daisy`

(library`cluster`

)- contains seven distance measures, but the function more than for practical use is for a demonstration of the script (for larger matrices, the calculation takes rather long).`distance`

(library`ecodist`

)

en/similarity_r.txt · Last modified: 2021/03/03 19:48 by David Zelený