R/plotEstimates.R

In MatchingFrontier: Computation of the Balance - Sample Size Frontier in Matching Methods for Causal Inference

plotEstimates <-
function(estimates.object,
         xlab = 'Number of Observations Pruned',
         ylab = 'Estimate',
         main = 'Effects Plot',
         xlim = NULL,
         ylim = NULL,
         mod.dependence.col = rgb(255,0,0,127, maxColorValue=255),
         mod.dependence.border.col = rgb(255,0,0,200, maxColorValue=255),
         line.col = rgb(102,0,0,255, maxColorValue=255),
         ...){

    if(is.null(xlim)){
        xlim <- c(0, max(estimates.object$Xs, na.rm = TRUE))
    }
    if(is.null(ylim)){
        ylim <- c(min(estimates.object$coefs - estimates.object$mod.dependence, na.rm = TRUE),
                  max(estimates.object$coefs + estimates.object$mod.dependence, na.rm = TRUE))
    }

    # remove NAs

    
    
    plot(1, type="n", main, xlab=xlab, ylab=ylab,
         xlim = xlim,
         ylim = ylim,
         ...)
    
    keep <- which(!is.na(estimates.object$mod.dependence))
    estimates.object$Xs <- estimates.object$Xs[keep]
    estimates.object$coefs <- estimates.object$coefs[keep]
    estimates.object$mod.dependence <- estimates.object$mod.dependence[keep]
    
    x0 <- rev(estimates.object$Xs)
    x1 <- estimates.object$Xs
    y0 <- rev(estimates.object$coefs + estimates.object$mod.dependence)
    y1 <- estimates.object$coefs - estimates.object$mod.dependence

    remove <- (is.na(x0) | is.na(x1) | is.na(y0) | is.na(y1))
    polygon(c(x0, x1),
            c(y0, y1),
            col = mod.dependence.col,
            border = mod.dependence.border.col)
    lines(estimates.object$Xs, estimates.object$coefs, type = 'l', col = line.col)
}