In jatotterdell/optimum: Package containing functions and scripts used for optimum study simulations

knitr::opts_chunk$set(
    echo = TRUE,
    message = FALSE,
    warning = FALSE,
    collapse = TRUE,
    comment = "#>"
)

library(optimum)
library(data.table)

Basic usage

simdat <- lapply(1:10, sim_trial_dat, p1tru = 0.2, p2tru = 0.1)
aggdat <- lapply(simdat, agg_trial_dat, stage_n = c(25, 50, 75, 100))
prbdat <- lapply(aggdat, est_trial_prob, ppos_q = c(0.95, 0.975))
decdat1 <- rbindlist(lapply(prbdat, dec_trial, sup_k = 0.95), idcol = "sim_id")
decdat2 <- rbindlist(lapply(prbdat, dec_trial, sup_k = 0.975), idcol = "sim_id")

Example

head(simdat[[10]])
head(prbdat[[10]])
head(decdat1)

Check trial decisions

dcast(decdat1[, .N, by = .(res, fin)], res ~ fin, value.var = "N", fill = 0)
dcast(decdat2[, .N, by = .(res, fin)], res ~ fin, value.var = "N", fill = 0)

Look at estimates when trial was stopped

dat_at_dec1 <- rbindlist(aggdat, idcol = "sim_id")[, 
  .(sim_id,resp,n1,n2,y1,y2,m1,m2,w1,w2,l1,l2,z1,z2,a1,b1,a2,b2)][
    decdat1, 
  on = .(sim_id, resp)]
dat_at_dec2 <- rbindlist(aggdat, idcol = "sim_id")[, 
  .(sim_id,resp,n1,n2,y1,y2,m1,m2,w1,w2,l1,l2,z1,z2,a1,b1,a2,b2)][
    decdat2, 
  on = .(sim_id, resp)]

dat_at_dec1[, .(mean1 = (a1 + w1) / (a1 + b1 + m1), 
                mean2 = (a2 + w2) / (a2 + b2 + m2))][, .(mean(mean1), mean(mean2))]
dat_at_dec2[, .(mean1 = (a1 + w1) / (a1 + b1 + m1), 
                mean2 = (a2 + w2) / (a2 + b2 + m2))][, .(mean(mean1), mean(mean2))]

jatotterdell/optimum documentation built on May 29, 2019, 1:24 p.m.