SimMeasure: Generation of Similarity Matrix from Sparse/ Low Information Content Data
Description
Returns a adjacency matrix with the similarity scores between individuals. If
threshold is provided, values where the absolute value of the
observation is less than the threshold are not considered. The similarity
measure is based on the percent difference between the observations. Details
of the algorithm can be found in the accompanying paper (see references).
Usage
SimMeasure(data, threshold=NULL, ...)
Arguments
data
Matrix object containing observation data on which to calculate
the similarity score.
threshold
The threshold value. Responses less than this value (absolute
value considered)are not used in calculating the similarity score.
...
Other parameters.
Value
SimMeasure returns an adjacency matrix containing edges corresponding
to the similarity of the observed values.
Details
Data can contain NA but may not contain NULL values. This method is designed
for datasets with high numbers of missing or uninformative values that can
be removed by setting the threshold value. Note that the
threshold value must be the same for all numbers.