sim_jaccard_emd: Simulate random data removal via Earth Mover's Distance...

In MrMaximumMax/FBCanalysis: Develop and evaluate time series data models based on fluctuation based clustering

Simulate random data removal via Earth Mover's Distance Cognate Cluster cluster assignment approach

Description

Simulate random data removal for a removal amount with indicated number of simulations from time series data list and determine Jaccard index for all clusters via Earth Mover's distance cognate cluster assignment approach.

Usage

sim_jaccard_emd(
  plist,
  parameter,
  removal,
  n_simu,
  method,
  n_clust,
  maxIter,
  normalize
)

Arguments

plist	Object of type list storing patient time series data (also see function: patient_list)
parameter	Parameter of interest in time series data list
removal	Amount of random data removal to determine Jaccard index
n_simu	Number of simulations
method	Clustering method (also see function: clust_matrix)
n_clust	Number of clusters (also see function: clust_matrix)
maxIter	Maximum iterations to determine Earth Mover's Distances (also see function: emd_matrix); default is 5,000 for this function
normalize	Indicates if parameter indicated needs to be normalized or not (TRUE by default)

Details

This method represents a novel approach and potential complementary method to sim_jaccard_global. First, clustering is performed on complete data without removal, serving as Gold Standard clusters. For every Gold Standard cluster then, all time series data from all patients is z-normalized and then assumed to be as one Gold Standard distribution. Subsequently, random data is removed form the time series data. Each leaky data distribution is then compared via Earth Mover's Distance to each Gold Standard Distribution. The Gold Standard cluster distribution to which the observed leaky distribution exhibits the lowest Earth Mover's Distance gets the assignment. This process is repeated until every leaky time series data distribution is assigned to a cluster. Afterwards, the Jaccard indices are calculated, comparing cluster members with complete and leaky data, for each cluster.

Value

Object of type matrix storing received Jaccard indices for indicated amount of random data removal for all clusters

References

Yossi Rubner, Carlo Tomasi, and Leonidas J Guibas. A metric for distributions with applications to image databases. In Sixth International Conference on Computer Vision (IEEE Cat. No. 98CH36271), pages 59–66. IEEE, 1998.

Examples

list <- patient_list(
"https://raw.githubusercontent.com/MrMaximumMax/FBCanalysis/master/demo/phys/data.csv",
GitHub = TRUE)
output <- sim_jaccard_emd(list, "PEF", 0.05, 10, "hierarchical", 2, 100)

MrMaximumMax/FBCanalysis documentation built on June 23, 2022, 8:21 p.m.