PAMA (version 1.2.0)

PAMA.F: This function implements Maximum Likelihood estimation of PAMA model.

Description

This function implements Maximum Likelihood estimation of PAMA model.

Usage

PAMA.F(datfile, nRe, threshold, iter = 1000, init = "EMM")

Arguments

datfile

A matrix or dataframe. This is the data where our algorithm will work on. Each row denotes a ranker's ranking. The data should be in entity-based format.

nRe

A number. Number of relevant entities.

threshold

A number (positive). The stopping threshold in determining convergence of MLE. If the two consecutive iterations of log-likelihood is smaller than the threshold, then the convergence is satisfied.

iter

A number. Numner of iterations of MCMC.

init

A string. This indicates which method is used to initiate the starting point of the aggregated ranking list. "mean" uses the sample mean. "EMM" uses the method from R package 'ExtMallows'.

Value

List. It contains MLE of all the parameters and log-likelihood. We use an iterative procedure to find the MLEs, so there are several values for each parameter until convergence.

1. I.mat: samples of I

2. phi.mat: samples of phi

3. smlgamma.mat: samples of gamma

4. l.mat: samples of log-likelihood

References

Wanchuang Zhu, Yingkai Jiang, Jun S. Liu, Ke Deng (2021) Partition-Mallows Model and Its Inference for Rank Aggregation. Journal of the American Statistical Association

Examples

# NOT RUN {
a=NBANFL()
PAMA.F(a$NBA,nRe=10,threshold=0.1,iter=10)
# }