aggregated_similarities: Aggregation similarities for sparse 3d representations

Description Usage Arguments

Description

These are similarity counterparts of aggregated_distances, which see. Generally similarities are computed from their distance counterparts through s=1-d or s=1/(1+d), but this need not always be the case. See the source code for the exact computation.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
asim_centroid(x, y = NULL, vecs, ptrans = NULL, strans = NULL,
  by = c("primary", "secondary", "row", "column"), precompute = TRUE,
  dist_type = "cosine")

asim_semantic_min_sum(x, y = NULL, vecs, ptrans = NULL, strans = NULL,
  by = c("primary", "secondary", "row", "column"), precompute = TRUE,
  dist_type = "cosine")

asim_semantic_min_max(x, y = NULL, vecs, ptrans = NULL, strans = NULL,
  by = c("primary", "secondary", "row", "column"), precompute = TRUE,
  dist_type = "cosine")

asim_rwmd(x, y = NULL, vecs, ptrans = NULL, strans = NULL,
  by = c("primary", "secondary", "row", "column"), precompute = TRUE,
  dist_type = "cosine")

Arguments

x

sparse or dense objects supported by simdist. See simdist-package for description of supported types and related terminology.

y

sparse or dense objects supported by simdist. See simdist-package for description of supported types and related terminology.

vecs

Dense matrix with columns

ptrans

Primary and secondary transformations. Can be either a function, string or a numeric vector. When a function, it must take 3 arguments - an object supported by simdist distance measures, names of a dimension ("primary" or "secondary") and logical na.rm. When a string it must name a transformation function without the semantic prefix (norm_, scale_ or trans_); see link{transformations}. When numeric vector it specifies weights to scale along the corresponding dimension.

strans

Primary and secondary transformations. Can be either a function, string or a numeric vector. When a function, it must take 3 arguments - an object supported by simdist distance measures, names of a dimension ("primary" or "secondary") and logical na.rm. When a string it must name a transformation function without the semantic prefix (norm_, scale_ or trans_); see link{transformations}. When numeric vector it specifies weights to scale along the corresponding dimension.

by

Dimension along which to perform distance computation. For all supported data structures computation along primary dimension is more or as efficient than along the secondary dimension.

precompute

logical Weather to optimize the computation for speed and precompute individual distances. The computation is method specific bug generally should be TRUE (the default) unless memory usage is a concern.

dist_type

distance to use across individual vectors in vecs


vspinu/simdist documentation built on May 3, 2019, 7:09 p.m.