| 1 |
|
| 2 |
# |
| 3 |
# Bayesian linear regression example |
| 4 |
# |
| 5 |
# by Will Freyman |
| 6 |
# |
| 7 |
|
| 8 |
|
| 9 |
x_obs <- readDataDelimitedFile(file="x.csv", header=FALSE, delimiter=",")[1] |
| 10 |
y_obs <- readDataDelimitedFile(file="y.csv", header=FALSE, delimiter=",")[1] |
| 11 |
|
| 12 |
beta ~ dnNormal(0, 1) |
| 13 |
alpha ~ dnNormal(0, 1) |
| 14 |
sigma ~ dnExponential(1) |
| 15 |
|
| 16 |
for (i in 1:x_obs.size()) { |
| 17 |
mu_y[i] := (beta * x_obs[i]) + alpha |
| 18 |
y[i] ~ dnNormal(mu_y[i], sigma) |
| 19 |
y[i].clamp(y_obs[i]) |
| 20 |
} |
| 21 |
|
| 22 |
mymodel = model(beta) |
| 23 |
|
| 24 |
moves[1] = mvSlide(beta) |
| 25 |
moves[2] = mvSlide(alpha) |
| 26 |
moves[3] = mvSlide(sigma) |
| 27 |
|
| 28 |
monitors[1] = mnFile(alpha,beta,sigma,filename="linear_regression.log",printgen=10) |
| 29 |
# monitors[2] = mnModel("linear_regression.log",printgen=1000) |
| 30 |
# monitors[3] = mnScreen() |
| 31 |
|
| 32 |
mymcmc = mcmc(mymodel, moves, monitors) |
| 33 |
mymcmc.burnin(generations=1000,tuningInterval=100) |
| 34 |
mymcmc.run(100000) |
| 35 |
|
| 36 |
q() |