EMRaff: Runs EM Algorithm for Rafferty Method

In jordanaron22/PartiallyObservedHMM: HMM for Partiall Observed Three Diagnostic Test Data

Description

Runs EM using adapted forward-backward that calculates partially observed data for Rafferty method. Calculates probability of being a stayer, intiial vector, transition matrix, and classification matrix Uses the following functions: ForwardLinearRaff, BackwardLinearRaff, CalcLikelihoodMover, CalcLikelihoodStayer, CalcInitialLinRaff, CalcTransitionLin,Raff CalcClassificationLinRaff, and CalcStayerLin

Usage

EMRaff(
  data_pattern,
  freq_vec,
  epsilon,
  time_length,
  init,
  tran,
  class,
  pi_0,
  lambda,
  max2,
  max3
)

Arguments

freq_vec	frequency vector of patterns
epsilon	threshold for EM convergence, stops when likelihood percentage increase is below epsilon
time_length	Number of observations per individual
init	initial probabilities
tran	transition probabilities
class	classification probabilities
pi_0	probability of being stayer
lambda	vector of length two, first value is weight given to value at time t-1, second value is weight given to value at time t-2
max2	number of values the second test can take on
max3	number of values the third test can take on
data_array	data array of three combined tests, patternized

Value

list of initial parameters, estimated parameters and likelihood initial and estimated parameter entries have same format 1) initial 2) transition 3) classification 4) stayer probability

jordanaron22/PartiallyObservedHMM documentation built on May 21, 2020, 6:49 p.m.