Specify a general marginal distribution
distr() is used within the function add_integration() to specify marginal
distributions for the covariates, via a corresponding inverse CDF. It is also
used in predict.stan_nma() to specify a distribution for the baseline
response (intercept) when predicting absolute outcomes.
distr(qfun, ...)
qfun |
an inverse CDF, either as a function name or a string |
... |
parameters of the distribution as arguments to |
The function qfun should have a formal argument called p. This
restriction serves as a crude check for inverse CDFs (e.g. an error will be
given if dnorm is used instead of qnorm). If a user-written CDF is
supplied, it must have an argument p which takes a vector of
probabilities.
An object of class distr.
add_integration() where distr() is used to specify marginal
distributions for covariates to integrate over, and predict.stan_nma()
where distr() is used to specify a distribution on the baseline response.
## Specifying marginal distributions for integration
df <- data.frame(x1_mean = 2, x1_sd = 0.5, x2 = 0.8)
# Distribution parameters are evaluated in the context of the data frame
add_integration(df,
x1 = distr(qnorm, mean = x1_mean, sd = x1_sd),
x2 = distr(qbern, prob = x2),
cor = diag(2))Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.