| EStep {RMallow} | R Documentation |
The Expectation step of the EM algorithm.
Description
Assigns each ranking the probability that it belongs to each cluster, given current parameters.
Usage
EStep(R, r, p, lambda, G, N, C, all.dists = NULL)
Arguments
R |
Current cluster modal sequences. |
r |
The data of partial or full rankings. |
p |
The proportion of the data currently assigned to each cluster. |
lambda |
The lambda parameters from Mallow's model for each cluster. |
G |
Number of clusters, length(R). |
N |
Number of rows in the data. |
C |
Vector of normalizing coefficients for the clusters. |
all.dists |
For efficiency, provide all of the Kendall distances between each sequence and each cluster mode. |
Value
Matrix where output[i, j] represents the current probability that subject "i" belongs to cluster "j".
Author(s)
Erik Gregory
References
"Mixtures of distance-based models for ranking data". Thomas Brendan Murphy & Donal Martin. 1 April 2002. Computational Statistics & Data Analysis 41 (2003) 645-655.