sequential.normal.dp {bdpopt} | R Documentation |
Create an object representing a sequential normal decision problem. A single observation with a normal distribution is made at each stage. The parameter is a true population mean with a conjugate normal prior. Under the assumption of a known population standard deviation, the variance of the posterior distribution for the parameter does not depend on the observations as is known at each stage. This implies that the state is one-dimensional and equals the mean of the posterior distribution for the parameter at each stage.
sequential.normal.dp(n.stages, group.size, tau, sigma, stage.cost, final.cost, final.gain)
n.stages |
The number of stages of the sequential decision problem. |
group.size |
The sample size at each stage. The individal samples are combined into a group mean, which is the single observation at each stage. |
tau |
The standard deviation of the prior for the unknown population mean before the first stage. |
sigma |
The population standard deviation for a single individual. The
standard deviation for the group response is this value divided by the
square root of |
stage.cost |
The cost of proceeding to the next stage. |
final.cost |
The cost payed at the final stage if a finalisation decision is taken (if a stopping decision is taken, this cost is not payed). |
final.gain |
A constant which is multiplied with the true population mean in order to obtain the utility gain at the final stage, if a finalisation decision is taken (if a stopping decision is taken, this gain is not included in the total utility). |
In all stages but the last, the two decisions available are to either continue and pay the stage cost or to stop and abort (which costs nothing). At the final stage, the two decisions available are to either finalise the process and obtain the final gain and pay the final cost or to stop and abort (whith no gain and no cost).
A list representing a sequential decision problem object. See
sequential.dp
for further description of the components.
Sebastian Jobjörnsson jobjorns@chalmers.se