9e216da9ef
After go1.16, go will use module mode by default, even when the repository is checked out under GOPATH or in a one-off directory. Add go.mod, go.sum to keep this repo buildable without opting out of the module mode. > go mod init github.com/mmcgrana/gobyexample > go mod tidy > go mod vendor In module mode, the 'vendor' directory is special and its contents will be actively maintained by the go command. pygments aren't the dependency the go will know about, so it will delete the contents from vendor directory. Move it to `third_party` directory now. And, vendor the blackfriday package. Note: the tutorial contents are not affected by the change in go1.16 because all the examples in this tutorial ask users to run the go command with the explicit list of files to be compiled (e.g. `go run hello-world.go` or `go build command-line-arguments.go`). When the source list is provided, the go command does not have to compute the build list and whether it's running in GOPATH mode or module mode becomes irrelevant.
49 lines
1.4 KiB
Plaintext
49 lines
1.4 KiB
Plaintext
# lsat.jags example from classic-bugs examples in JAGS
|
|
# See http://sourceforge.net/projects/mcmc-jags/files/Examples/2.x/
|
|
var
|
|
response[R,T], m[R], culm[R], alpha[T], a[T], theta[N], r[N,T],
|
|
p[N,T], beta, theta.new, p.theta[T], p.item[R,T], P.theta[R];
|
|
data {
|
|
for (j in 1:culm[1]) {
|
|
r[j, ] <- response[1, ];
|
|
}
|
|
for (i in 2:R) {
|
|
for (j in (culm[i - 1] + 1):culm[i]) {
|
|
r[j, ] <- response[i, ];
|
|
}
|
|
}
|
|
}
|
|
model {
|
|
# 2-parameter Rasch model
|
|
for (j in 1:N) {
|
|
for (k in 1:T) {
|
|
probit(p[j,k]) <- delta[k]*theta[j] - eta[k];
|
|
r[j,k] ~ dbern(p[j,k]);
|
|
}
|
|
theta[j] ~ dnorm(0,1);
|
|
}
|
|
|
|
# Priors
|
|
for (k in 1:T) {
|
|
eta[k] ~ dnorm(0,0.0001);
|
|
e[k] <- eta[k] - mean(eta[]); # sum-to-zero constraint
|
|
|
|
delta[k] ~ dnorm(0,1) T(0,); # constrain variance to 1, slope +ve
|
|
d[k] <- delta[k]/pow(prod(delta), 1/T); # PRODUCT_k (d_k) = 1
|
|
|
|
g[k] <- e[k]/d[k]; # equivalent to B&A's threshold parameters
|
|
}
|
|
|
|
# Compute probability of response pattern i, for later use in computing G^2
|
|
theta.new ~ dnorm(0,1); # ability parameter for random student
|
|
for(k in 1:T) {
|
|
probit(p.theta[k]) <- delta[k]*theta.new - eta[k];
|
|
for(i in 1:R) {
|
|
p.item[i,k] <- p.theta[k]^response[i,k] * (1-p.theta[k])^(1-response[i,k]);
|
|
}
|
|
}
|
|
for(i in 1:R) {
|
|
P.theta[i] <- prod(p.item[i,])
|
|
}
|
|
}
|