simulations/summarize_estimator.R

   1
   2 summarize.estimator <- function(df, suffix='naive', coefname='x'){
   3
   4     part <- df[,c('N',
   5                   'm',
   6                   'Bxy',
   7                   paste0('B',coefname,'y.est.',suffix),
   8                   paste0('B',coefname,'y.ci.lower.',suffix),
   9                   paste0('B',coefname,'y.ci.upper.',suffix),
  10                   'y_explained_variance',
  11                   'Bzx',
  12                   'Bzy',
  13                   'accuracy_imbalance_difference'
  14                   ),
  15                with=FALSE]
  16
  17
  18     true.in.ci <- as.integer((part$Bxy >= part[[paste0('B',coefname,'y.ci.lower.',suffix)]]) & (part$Bxy <= part[[paste0('B',coefname,'y.ci.upper.',suffix)]]))
  19     zero.in.ci <- as.integer(0 >= part[[paste0('B',coefname,'y.ci.lower.',suffix)]]) & (0 <= part[[paste0('B',coefname,'y.ci.upper.',suffix)]])
  20     bias <- part[[paste0('B',coefname,'y')]] - part[[paste0('B',coefname,'y.est.',suffix)]]
  21     sign.correct <- as.integer(sign(part$Bxy) == sign(part[[paste0('B',coefname,'y.est.',suffix)]]))
  22
  23     part <- part[,':='(true.in.ci = true.in.ci,
  24                        zero.in.ci = zero.in.ci,
  25                        bias=bias,
  26                        sign.correct =sign.correct)]
  27
  28     part.plot <- part[, .(p.true.in.ci = mean(true.in.ci),
  29                           mean.bias = mean(bias),
  30                           mean.est = mean(.SD[[paste0('B',coefname,'y.est.',suffix)]]),
  31                           var.est = var(.SD[[paste0('B',coefname,'y.est.',suffix)]]),
  32                           est.upper.95 = quantile(.SD[[paste0('B',coefname,'y.est.',suffix)]],0.975,na.rm=T),
  33                           est.lower.95 = quantile(.SD[[paste0('B',coefname,'y.est.',suffix)]],0.025,na.rm=T),
  34                           mean.ci.upper = mean(.SD[[paste0('B',coefname,'y.ci.upper.',suffix)]],na.rm=T),
  35                           mean.ci.lower = mean(.SD[[paste0('B',coefname,'y.ci.lower.',suffix)]],na.rm=T),
  36                           ci.upper.975 = quantile(.SD[[paste0('B',coefname,'y.ci.upper.',suffix)]],0.975,na.rm=T),
  37                           ci.upper.025 = quantile(.SD[[paste0('B',coefname,'y.ci.upper.',suffix)]],0.025,na.rm=T),
  38                           ci.lower.975 = quantile(.SD[[paste0('B',coefname,'y.ci.lower.',suffix)]],0.975,na.rm=T),
  39                           ci.lower.025 = quantile(.SD[[paste0('B',coefname,'y.ci.lower.',suffix)]],0.025,na.rm=T),
  40                           N.ci.is.NA = sum(is.na(.SD[[paste0('B',coefname,'y.ci.lower.',suffix)]])),
  41                           N.sims = .N,
  42                           p.sign.correct = mean(as.integer(sign.correct & (! zero.in.ci))),
  43                           variable=coefname,
  44                           method=suffix
  45                           ),
  46                       by=c("N","m",'y_explained_variance','Bzx', 'Bzy', 'accuracy_imbalance_difference')
  47                       ]
  48
  49     return(part.plot)
  50 }