simulations/plot_irr_example.R

   1 source("RemembR/R/RemembeR.R")
   2 library(arrow)
   3 library(data.table)
   4 library(ggplot2)
   5 library(filelock)
   6 library(argparser)
   7
   8 parser <- arg_parser("Simulate data and fit corrected models.")
   9 parser <- add_argument(parser, "--infile", default="", help="name of the file to read.")
  10 parser <- add_argument(parser, "--name", default="", help="The name to safe the data to in the remember file.")
  11 args <- parse_args(parser)
  12 source("summarize_estimator.R")
  13
  14 build_plot_dataset <- function(df){
  15
  16     x.true <-  summarize.estimator(df, 'true','x')
  17
  18     z.true <-  summarize.estimator(df, 'true','z')
  19
  20     x.naive <-  summarize.estimator(df, 'naive','x')
  21
  22     z.naive <-  summarize.estimator(df, 'naive','z')
  23
  24     x.loa0.feasible <- summarize.estimator(df, 'loa0.feasible','x')
  25
  26     z.loa0.feasible <- summarize.estimator(df,'loa0.feasible','z')
  27
  28     x.loa0.mle <- summarize.estimator(df, 'loa0.mle', 'x')
  29
  30     z.loa0.mle <- summarize.estimator(df, 'loa0.mle', 'z')
  31
  32     x.loco.feasible <- summarize.estimator(df, 'loco.feasible', 'x')
  33
  34     z.loco.feasible <- summarize.estimator(df, 'loco.feasible', 'z')
  35
  36     x.loco.mle <- summarize.estimator(df, 'loco.mle', 'x')
  37
  38     z.loco.mle <- summarize.estimator(df, 'loco.mle', 'z')
  39
  40     x.loco.mle <- summarize.estimator(df, 'loco.mle', 'x')
  41
  42     z.loco.amelia <- summarize.estimator(df, 'amelia.full', 'z')
  43     x.loco.amelia <- summarize.estimator(df, 'amelia.full', 'x')
  44
  45     z.loco.zhang <- summarize.estimator(df, 'zhang', 'z')
  46     x.loco.zhang <- summarize.estimator(df, 'zhang', 'x')
  47
  48
  49     z.loco.gmm <- summarize.estimator(df, 'gmm', 'z')
  50     x.loco.gmm <- summarize.estimator(df, 'gmm', 'x')
  51
  52
  53
  54
  55     ## x.mle <- summarize.estimator(df, 'mle', 'x')
  56
  57     ## z.mle <- summarize.estimator(df, 'mle', 'z')
  58
  59     accuracy <- df[,mean(accuracy)]
  60     plot.df <- rbindlist(list(x.true,z.true,x.loa0.feasible,z.loa0.feasible,x.loa0.mle,z.loa0.mle,x.loco.feasible, z.loco.feasible, x.loco.mle, z.loco.mle, x.loco.amelia, z.loco.amelia,x.loco.zhang, z.loco.zhang,x.loco.gmm, z.loco.gmm,x.naive,z.naive),use.names=T)
  61     plot.df[,accuracy := accuracy]
  62     plot.df <- plot.df[,":="(sd.est=sqrt(var.est)/N.sims)]
  63     return(plot.df)
  64 }
  65
  66
  67 sims.df <- read_feather(args$infile)
  68 print(unique(sims.df$N))
  69
  70 # df <- df[apply(df,1,function(x) !any(is.na(x)))]
  71
  72 if(!('Bzx' %in% names(sims.df)))
  73     sims.df[,Bzx:=NA]
  74
  75 if(!('accuracy_imbalance_difference' %in% names(sims.df)))
  76     sims.df[,accuracy_imbalance_difference:=NA]
  77
  78 unique(sims.df[,'accuracy_imbalance_difference'])
  79
  80 #plot.df <- build_plot_dataset(df[accuracy_imbalance_difference==0.1][N==700])
  81 plot.df <- build_plot_dataset(sims.df)
  82 change.remember.file("remember_irr.RDS",clear=TRUE)
  83 remember(plot.df,args$name)
  84
  85
  86 set.remember.prefix(gsub("plot.df.","",args$name))
  87 remember(median(sims.df$loco.accuracy),'med.loco.acc')
  88
  89 #ggplot(df,aes(x=Bxy.est.mle)) + geom_histogram() + facet_grid(accuracy_imbalance_difference ~ Bzy)
  90
  91 ## ## ## df[gmm.ER_pval<0.05]
  92
  93 ## plot.df.test <- plot.df[,':='(method=factor(method,levels=c("Naive","Multiple imputation", "Multiple imputation (Classifier features unobserved)","Regression Calibration","2SLS+gmm","Bespoke MLE", "Feasible"),ordered=T),
  94 ##                                    N=factor(N),
  95 ##                                    m=factor(m))]
  96
  97 ## plot.df.test <- plot.df.test[(variable=='x') & (method!="Multiple imputation (Classifier features unobserved)")]
  98 ## p <- ggplot(plot.df.test, aes(y=mean.est, ymax=mean.est + var.est/2, ymin=mean.est-var.est/2, x=method))
  99 ## p <- p + geom_hline(data=plot.df.test, mapping=aes(yintercept=0.1),linetype=2)
 100
 101 ## p <- p + geom_pointrange() + facet_grid(N~m,as.table=F,scales='free') + scale_x_discrete(labels=label_wrap_gen(4))
 102 ## print(p)
 103
 104 ## plot.df.test <- plot.df[,':='(method=factor(method,levels=c("Naive","Multiple imputation", "Multiple imputation (Classifier features unobserved)","Regression Calibration","2SLS+gmm","Bespoke MLE", "Feasible"),ordered=T),
 105 ##                                    N=factor(N),
 106 ##                                    m=factor(m))]
 107
 108 ## plot.df.test <- plot.df.test[(variable=='z') & (method!="Multiple imputation (Classifier features unobserved)")]
 109 ## p <- ggplot(plot.df.test, aes(y=mean.est, ymax=mean.est + var.est/2, ymin=mean.est-var.est/2, x=method))
 110 ## p <- p + geom_hline(data=plot.df.test, mapping=aes(yintercept=-0.1),linetype=2)
 111
 112 ## p <- p + geom_pointrange() + facet_grid(m~N,as.table=F,scales='free') + scale_x_discrete(labels=label_wrap_gen(4))
 113 ## print(p)
 114
 115
 116 ## x.mle <- df[,.(N,m,Bxy.est.mle,Bxy.ci.lower.mle, Bxy.ci.upper.mle, y_explained_variance, Bzx, Bzy, accuracy_imbalance_difference)]
 117 ## x.mle.plot <- x.mle[,.(mean.est = mean(Bxy.est.mle),
 118 ##                        var.est = var(Bxy.est.mle),
 119 ##                        N.sims = .N,
 120 ##                        variable='z',
 121 ##                        method='Bespoke MLE'
 122 ##                        ),
 123 ##                     by=c("N","m",'y_explained_variance', 'Bzx', 'Bzy','accuracy_imbalance_difference')]
 124
 125 ## z.mle <- df[,.(N,m,Bzy.est.mle,Bzy.ci.lower.mle, Bzy.ci.upper.mle, y_explained_variance, Bzx, Bzy, accuracy_imbalance_difference)]
 126
 127 ## z.mle.plot <- z.mle[,.(mean.est = mean(Bzy.est.mle),
 128 ##                        var.est = var(Bzy.est.mle),
 129 ##                        N.sims = .N,
 130 ##                        variable='z',
 131 ##                        method='Bespoke MLE'
 132 ##                        ),
 133 ##                     by=c("N","m",'y_explained_variance','Bzx')]
 134
 135 ## plot.df <- z.mle.plot
 136 ## plot.df.test <- plot.df[,':='(method=factor(method,levels=c("Naive","Multiple imputation", "Multiple imputation (Classifier features unobserved)","Regression Calibration","2SLS+gmm","Bespoke MLE", "Feasible"),ordered=T),
 137 ##                                    N=factor(N),
 138 ##                                    m=factor(m))]
 139
 140 ## plot.df.test <- plot.df.test[(variable=='z') & (m != 1000) & (m!=500) & (method!="Multiple imputation (Classifier features unobserved)")]
 141 ## p <- ggplot(plot.df.test, aes(y=mean.est, ymax=mean.est + var.est/2, ymin=mean.est-var.est/2, x=method))
 142 ## p <- p + geom_hline(aes(yintercept=0.2),linetype=2)
 143
 144 ## p <- p + geom_pointrange() + facet_grid(m~Bzx, Bzy,as.table=F) + scale_x_discrete(labels=label_wrap_gen(4))
 145 ## print(p)
 146
 147
 148 ## ## ggplot(plot.df[variable=='x'], aes(y=mean.est, ymax=mean.est + var.est/2, ymin=mean.est-var.est/2, x=method)) + geom_pointrange() + facet_grid(-m~N) + scale_x_discrete(labels=label_wrap_gen(10))
 149
 150 ## ## ggplot(plot.df,aes(y=N,x=m,color=p.sign.correct)) + geom_point() + facet_grid(variable ~ method) + scale_color_viridis_c(option='D') + theme_minimal() + xlab("Number of gold standard labels") + ylab("Total sample size")
 151
 152 ## ## ggplot(plot.df,aes(y=N,x=m,color=abs(mean.bias))) + geom_point() + facet_grid(variable ~ method) + scale_color_viridis_c(option='D') + theme_minimal() + xlab("Number of gold standard labels") + ylab("Total sample size")