Estimation using mlogit in R -


i have paper author proposes d-optimal design nested multinomial logit model no-choice. design follows:

cs 1:

alt a: 1 2 2 1 2 2

alt b: 2 2 1 2 1 2

alt c: 0 0 0 0 0 0

cs 2:

alt 1: 2 1 1 2 2 2

alt 2: 2 2 2 2 1 1

alt 3: 0 0 0 0 0 0

cs 3:

alt 1: 2 2 1 1 1 2

alt 2: 1 1 2 2 2 1

alt 3: 0 0 0 0 0 0

cs 4:

alt 1: 2 2 1 1 2 1

alt 2: 1 1 2 2 1 2

alt 3: 0 0 0 0 0 0

cs 5:

alt 1: 1 2 1 2 2 1

alt 2: 2 1 2 1 1 2

alt 3: 0 0 0 0 0 0

cs 6:

alt 1: 1 2 1 2 1 2

alt 2: 2 1 2 1 2 1

alt 3: 0 0 0 0 0 0

cs 7:

alt 1: 2 1 1 2 2 1

alt 2: 1 2 2 1 2 2

alt 3: 0 0 0 0 0 0

cs 8:

alt 1: 2 1 2 2 2 1

alt 2: 1 2 2 1 1 2

alt 3: 0 0 0 0 0 0

the design d-optimal design 8 choice sets , every choice set has 2 alternatives (6 two-level attributes alternative). tried estimate nested multinomial logit model in r using mlogit , design. thus, made simulated data using random response follow:

individual--choiceid--mode----choice----aa--ab--ac--ad--ae--af
------1-------------1----------a-------false-----1-----2-----2-----1-----2-----2
------1-------------1----------b-------false-----2-----2-----1-----2-----1-----2
------1-------------1----------c--------true-----0-----0-----0-----0-----0-----0
------1-------------2----------a-------false-----2-----1-----1-----2-----2-----2
------1-------------2----------b--------true-----2-----2-----2-----2-----1-----1
------1-------------2----------c-------false-----0-----0-----0-----0-----0-----0
------1-------------3----------a-------false-----2-----2-----1-----1-----1-----2
------1-------------3----------b-------false-----1-----1-----2-----2-----2-----1

the data contains responses of 30 people, means, data contains 240 choices (8 choices person). when used command

sim = read.csv("sim.csv",header=t)   dat <- mlogit.data(sim, choice = "choice", shape = "long",                           alt.var = "mode", chid.var = "choideid",id.var="individual") nl.dat <- mlogit(choice ~ aa+ab+ac+ad+ae, dat, reflevel="c",                        nests = list(our = c("a","b") , sq = c("c")),                        unscaled = true) 

i got error

error in solve.default(crossprod(attr(x, "gradi")[, !fixed])) : system computationally singular: reciprocal condition number = 1.13304e-18

someone has idea i'm doing wrong?


Comments

Popular posts from this blog

angularjs - ADAL JS Angular- WebAPI add a new role claim to the token -

php - CakePHP HttpSockets send array of paramms -

node.js - Using Node without global install -