simulations/plot_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
  13 summarize.estimator <- function(df, suffix='naive', coefname='x'){
  14
  15     part <- df[,c('N',
  16                   'm',
  17                   'Bxy',
  18                   paste0('B',coefname,'y.est.',suffix),
  19                   paste0('B',coefname,'y.ci.lower.',suffix),
  20                   paste0('B',coefname,'y.ci.upper.',suffix),
  21                   'y_explained_variance',
  22                   'Bzx',
  23                   'Bzy',
  24                   'accuracy_imbalance_difference'
  25                   ),
  26                with=FALSE]
  27
  28     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)]]))
  29     zero.in.ci <- as.integer(0 >= part[[paste0('B',coefname,'y.ci.lower.',suffix)]]) & (0 <= part[[paste0('B',coefname,'y.ci.upper.',suffix)]])
  30     bias <- part$Bxy - part[[paste0('B',coefname,'y.est.',suffix)]]
  31     sign.correct <- as.integer(sign(part$Bxy) == sign(part[[paste0('B',coefname,'y.est.',suffix)]]))
  32
  33     part <- part[,':='(true.in.ci = true.in.ci,
  34                        zero.in.ci = zero.in.ci,
  35                        bias=bias,
  36                        sign.correct =sign.correct)]
  37
  38     part.plot <- part[, .(p.true.in.ci = mean(true.in.ci),
  39                           mean.bias = mean(bias),
  40                           mean.est = mean(.SD[[paste0('B',coefname,'y.est.',suffix)]]),
  41                           var.est = var(.SD[[paste0('B',coefname,'y.est.',suffix)]]),
  42                           est.upper.95 = quantile(.SD[[paste0('B',coefname,'y.est.',suffix)]],0.95,na.rm=T),
  43                           est.lower.95 = quantile(.SD[[paste0('B',coefname,'y.est.',suffix)]],0.05,na.rm=T),
  44                           N.sims = .N,
  45                           p.sign.correct = mean(as.integer(sign.correct & (! zero.in.ci))),
  46                           variable=coefname,
  47                           method=suffix
  48                           ),
  49                       by=c("N","m",'y_explained_variance','Bzx', 'Bzy', 'accuracy_imbalance_difference')
  50                       ]
  51
  52     return(part.plot)
  53 }
  54
  55 build_plot_dataset <- function(df){
  56
  57     x.true <-  summarize.estimator(df, 'true','x')
  58
  59     z.true <-  summarize.estimator(df, 'true','z')
  60
  61     x.naive <- summarize.estimator(df, 'naive','x')
  62
  63     z.naive <- summarize.estimator(df,'naive','z')
  64
  65     x.feasible <- summarize.estimator(df, 'feasible', 'x')
  66
  67     z.feasible <- summarize.estimator(df, 'feasible', 'z')
  68
  69     x.amelia.full <- summarize.estimator(df, 'amelia.full', 'x')
  70
  71     z.amelia.full <- summarize.estimator(df, 'amelia.full', 'z')
  72
  73     x.mecor <- summarize.estimator(df, 'mecor', 'x')
  74
  75     z.mecor <- summarize.estimator(df, 'mecor', 'z')
  76
  77     x.mecor <- summarize.estimator(df, 'mecor', 'x')
  78
  79     z.mecor <- summarize.estimator(df, 'mecor', 'z')
  80
  81     x.mle <- summarize.estimator(df, 'mle', 'x')
  82
  83     z.mle <- summarize.estimator(df, 'mle', 'z')
  84
  85     x.zhang <- summarize.estimator(df, 'zhang', 'x')
  86
  87     z.zhang <- summarize.estimator(df, 'zhang', 'z')
  88
  89     x.gmm <- summarize.estimator(df, 'gmm', 'x')
  90
  91     z.gmm <- summarize.estimator(df, 'gmm', 'z')
  92
  93     accuracy <- df[,mean(accuracy)]
  94     plot.df <- rbindlist(list(x.true,z.true,x.naive,z.naive,x.amelia.full,z.amelia.full,x.mecor, z.mecor, x.gmm, z.gmm, x.feasible, z.feasible,z.mle, x.mle, x.zhang, z.zhang, x.gmm, z.gmm),use.names=T)
  95     plot.df[,accuracy := accuracy]
  96     plot.df <- plot.df[,":="(sd.est=sqrt(var.est)/N.sims)]
  97     return(plot.df)
  98 }
  99
 100
 101 plot.df <- read_feather(args$infile)
 102 print(unique(plot.df$N))
 103
 104 # df <- df[apply(df,1,function(x) !any(is.na(x)))]
 105
 106 if(!('Bzx' %in% names(plot.df)))
 107     plot.df[,Bzx:=NA]
 108
 109 if(!('accuracy_imbalance_difference' %in% names(plot.df)))
 110     plot.df[,accuracy_imbalance_difference:=NA]
 111
 112 unique(plot.df[,'accuracy_imbalance_difference'])
 113
 114 #plot.df <- build_plot_dataset(df[accuracy_imbalance_difference==0.1][N==700])
 115 plot.df <- build_plot_dataset(plot.df)
 116
 117 remember(plot.df,args$name)
 118
 119 #ggplot(df,aes(x=Bxy.est.mle)) + geom_histogram() + facet_grid(accuracy_imbalance_difference ~ Bzy)
 120
 121 ## ## ## df[gmm.ER_pval<0.05]
 122
 123 ## 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),
 124 ##                                    N=factor(N),
 125 ##                                    m=factor(m))]
 126
 127 ## plot.df.test <- plot.df.test[(variable=='x') & (method!="Multiple imputation (Classifier features unobserved)")]
 128 ## p <- ggplot(plot.df.test, aes(y=mean.est, ymax=mean.est + var.est/2, ymin=mean.est-var.est/2, x=method))
 129 ## p <- p + geom_hline(data=plot.df.test, mapping=aes(yintercept=0.1),linetype=2)
 130
 131 ## p <- p + geom_pointrange() + facet_grid(N~m,as.table=F,scales='free') + scale_x_discrete(labels=label_wrap_gen(4))
 132 ## print(p)
 133
 134 ## 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),
 135 ##                                    N=factor(N),
 136 ##                                    m=factor(m))]
 137
 138 ## plot.df.test <- plot.df.test[(variable=='z') & (method!="Multiple imputation (Classifier features unobserved)")]
 139 ## p <- ggplot(plot.df.test, aes(y=mean.est, ymax=mean.est + var.est/2, ymin=mean.est-var.est/2, x=method))
 140 ## p <- p + geom_hline(data=plot.df.test, mapping=aes(yintercept=-0.1),linetype=2)
 141
 142 ## p <- p + geom_pointrange() + facet_grid(m~N,as.table=F,scales='free') + scale_x_discrete(labels=label_wrap_gen(4))
 143 ## print(p)
 144
 145
 146 ## x.mle <- df[,.(N,m,Bxy.est.mle,Bxy.ci.lower.mle, Bxy.ci.upper.mle, y_explained_variance, Bzx, Bzy, accuracy_imbalance_difference)]
 147 ## x.mle.plot <- x.mle[,.(mean.est = mean(Bxy.est.mle),
 148 ##                        var.est = var(Bxy.est.mle),
 149 ##                        N.sims = .N,
 150 ##                        variable='z',
 151 ##                        method='Bespoke MLE'
 152 ##                        ),
 153 ##                     by=c("N","m",'y_explained_variance', 'Bzx', 'Bzy','accuracy_imbalance_difference')]
 154
 155 ## z.mle <- df[,.(N,m,Bzy.est.mle,Bzy.ci.lower.mle, Bzy.ci.upper.mle, y_explained_variance, Bzx, Bzy, accuracy_imbalance_difference)]
 156
 157 ## z.mle.plot <- z.mle[,.(mean.est = mean(Bzy.est.mle),
 158 ##                        var.est = var(Bzy.est.mle),
 159 ##                        N.sims = .N,
 160 ##                        variable='z',
 161 ##                        method='Bespoke MLE'
 162 ##                        ),
 163 ##                     by=c("N","m",'y_explained_variance','Bzx')]
 164
 165 ## plot.df <- z.mle.plot
 166 ## 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),
 167 ##                                    N=factor(N),
 168 ##                                    m=factor(m))]
 169
 170 ## plot.df.test <- plot.df.test[(variable=='z') & (m != 1000) & (m!=500) & (method!="Multiple imputation (Classifier features unobserved)")]
 171 ## p <- ggplot(plot.df.test, aes(y=mean.est, ymax=mean.est + var.est/2, ymin=mean.est-var.est/2, x=method))
 172 ## p <- p + geom_hline(aes(yintercept=0.2),linetype=2)
 173
 174 ## p <- p + geom_pointrange() + facet_grid(m~Bzx, Bzy,as.table=F) + scale_x_discrete(labels=label_wrap_gen(4))
 175 ## print(p)
 176
 177
 178 ## ## 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))
 179
 180 ## ## 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")
 181
 182 ## ## 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")