在 R 中排除后重新出现异常值

如何解决在 R 中排除后重新出现异常值

所以我正在为我的论文分析一些有关细胞培养的数据，我正在研究细胞形态。关于细胞表面积，我通过简单的子集化排除了一些异常值。下面是一些有代表性的代码：

function SetMemberListFilterFromCookie() {
    if ($(".MemberListPage").length > 0) {
        var cookieValue = readCookie("searchFilterCookie"));
        if (cookieValue) {
            getMembers(cookieValue); 

            var array1 = cookieValue.split("&");
            var array2 = [];
            for (var i = 0; i <= array1.length - 1; i++) {
                array2.push(array1[i].split("="));
            }
            // This section seems to be an issue in CHROME
            // CheckBoxes are not always checked on page load
            for (var c = 0; c < array2.length; c++) {
                var id = array2[c][0] + "-" + array2[c][1];
                $("#" + id).prop("checked",true);
            }
            
            // checkBox panels are initially hidden on page
            // if any checkBoxes within a filterPanel have been checked,the filterPanel is opened.
            var filterPanels = $(".filter-panel ul");
            filterPanels.each(function() {
                var checked = $(this).find("input:checked");
                if (checked.length > 0) {
                    animateFilterSection($(this).prev());
                }
            });
        }
    }
}

function animateFilterSection(btn) {
    btn.next("ul").slidetoggle({
        duration: 100,complete: function() {
            btn.toggleClass("plus minus");
        }
    });
}

function readCookie(name) {
    var nameEQ = encodeURIComponent(name) + "=";
    var ca = document.cookie.split(";");
    for (var i = 0; i < ca.length; i++) {
        var c = ca[i];
        while (c.charat(0) === " ")
            c = c.substring(1,c.length);
        if (c.indexOf(nameEQ) === 0)
            return decodeURIComponent(c.substring(nameEQ.length,c.length));
    }
    return null;
}

library(ggpubr)
random <- data.frame(x = c(sample(1:20,40,replace = TRUE),30),y = c(sample(1:20,30,52)) 
Boxplot(random)

random2 <- subset(random,random < 21) 
Boxplot(random2) 

此后的箱线图（使用 compare_means(x~y,data=random2,method="wilcox.test",paired = FALSE,group.by = NULL,ref.group = NULL) Boxplotrandom <- ggBoxplot(random2,x = "x",y = "y") Boxplotrandom <- Boxplotrandom + stat_compare_means() Boxplotrandom ）看起来不错，但是当我执行 wilcoxon 有符号秩检验并使用 Boxplot(random2) 为结果制作箱线图时，我再次得到异常值。我怎样才能摆脱这些？

提前致谢。