在 R 中按 id 执行一个样本学生 t 检验答案

【问题标题】：performing One sample Student’s t-test by id in R在 R 中按 id 执行一个样本学生 t 检验
【发布时间】：2018-11-19 21:22:33
【问题描述】：

在我的数据集中，我有几个月的数据。

df=structure(list(id = c(1030879980L, 1030879990L), jan = c(170L, 
265L), feb = c(153L, 332L), march = c(170L, 290L), apr = c(1L, 
425L), may = c(66L, 406L), jume = c(125L, 352L), jul = c(129L, 
339L), aug = c(-109L, 470L), sept = c(56L, 486L), oct = c(37L, 
440L), nov = c(52L, 589L), dec = c(63L, 659L)), .Names = c("id", 
"jan", "feb", "march", "apr", "may", "jume", "jul", "aug", "sept", 
"oct", "nov", "dec"), class = "data.frame", row.names = c(NA, 
-2L))

我必须为每个 id 执行一个样本学生的 t 检验使用参考值

这里有参考值的数据

 ref=structure(list(jan = 507L, feb = 502L, march = 431L, apr = 429L, 
        may = 449L, jume = 368L, jul = 406L, aug = 290L, sept = 309L, 
        oct = 371L, nov = 481L, dec = 536L), .Names = c("jan", "feb", 
    "march", "apr", "may", "jume", "jul", "aug", "sept", "oct", "nov", 
    "dec"), class = "data.frame", row.names = c(NA, -1L))

所以我很简单

#the first id 1030879980

    a = c(170,153,170,1,66,125,129,-109,56,37,52,63)
    #jan reference values for january
    t.test (a, mu=507)
    #feb reference values for febrary
    t.test (a, mu=502)

但是我怎样才能按月为每个 id 执行呢？当然，当我（手动）这样做时，它会很长。有很多id。

【问题讨论】：

也许只是sapply(dat$month, function (x) t.test(a, mu=x))？你可能需要重塑或其他东西。或者 sapply(dat, function (x) t.test(a, mu=x)) 如果您的 data.frame 是单行引用。
@lmo，我做了
sapply(dat, function (x) t.test(dat, mu=x)) Rerun with Debug Error in t.test.default(dat, mu = x) : 'mu' must be一个数字
数据集的结构是每个月都有单独的列，我可以在可重现的例子中看到它
x 必须是参考数据集中的参考值

标签： r dataframe dplyr plyr

【解决方案1】：

您可以执行以下操作。将结果存储在结果列表中。遍历 id，存储 t.tests 并用 id 命名列表。

result <- vector("list", nrow(df))
for(i in seq_along(df$id)) {
  result[[i]] <- t(sapply(t(ref), function (x) t.test(df[i, -1], mu=x, data.name = i)))
}
names(result) <- df$id

result
$`1030879980`
      statistic parameter p.value      conf.int  estimate null.value alternative method              data.name  
 [1,] -18.52462 11        1.214017e-09 Numeric,2 76.08333 507        "two.sided" "One Sample t-test" "df[i, -1]"
 [2,] -18.30968 11        1.375191e-09 Numeric,2 76.08333 502        "two.sided" "One Sample t-test" "df[i, -1]"
 [3,] -15.25747 11        9.526113e-09 Numeric,2 76.08333 431        "two.sided" "One Sample t-test" "df[i, -1]"
 [4,] -15.17149 11        1.01107e-08  Numeric,2 76.08333 429        "two.sided" "One Sample t-test" "df[i, -1]"
 [5,] -16.03127 11        5.649016e-09 Numeric,2 76.08333 449        "two.sided" "One Sample t-test" "df[i, -1]"
 [6,] -12.54917 11        7.334351e-08 Numeric,2 76.08333 368        "two.sided" "One Sample t-test" "df[i, -1]"
 [7,] -14.18274 11        2.05238e-08  Numeric,2 76.08333 406        "two.sided" "One Sample t-test" "df[i, -1]"
 [8,] -9.196035 11        1.696865e-06 Numeric,2 76.08333 290        "two.sided" "One Sample t-test" "df[i, -1]"
 [9,] -10.01282 11        7.298909e-07 Numeric,2 76.08333 309        "two.sided" "One Sample t-test" "df[i, -1]"
[10,] -12.67813 11        6.598342e-08 Numeric,2 76.08333 371        "two.sided" "One Sample t-test" "df[i, -1]"
[11,] -17.40691 11        2.357586e-09 Numeric,2 76.08333 481        "two.sided" "One Sample t-test" "df[i, -1]"
[12,] -19.7713  11        6.045928e-10 Numeric,2 76.08333 536        "two.sided" "One Sample t-test" "df[i, -1]"

$`1030879990`
      statistic  parameter p.value     conf.int  estimate null.value alternative method              data.name  
 [1,] -2.523598  11        0.02829384  Numeric,2 421.0833 507        "two.sided" "One Sample t-test" "df[i, -1]"
 [2,] -2.376735  11        0.03671235  Numeric,2 421.0833 502        "two.sided" "One Sample t-test" "df[i, -1]"
 [3,] -0.2912785 11        0.7762574   Numeric,2 421.0833 431        "two.sided" "One Sample t-test" "df[i, -1]"
 [4,] -0.2325333 11        0.8203937   Numeric,2 421.0833 429        "two.sided" "One Sample t-test" "df[i, -1]"
 [5,] -0.8199858 11        0.4296343   Numeric,2 421.0833 449        "two.sided" "One Sample t-test" "df[i, -1]"
 [6,] 1.559197   11        0.1472385   Numeric,2 421.0833 368        "two.sided" "One Sample t-test" "df[i, -1]"
 [7,] 0.4430371  11        0.6663254   Numeric,2 421.0833 406        "two.sided" "One Sample t-test" "df[i, -1]"
 [8,] 3.850262   11        0.00269828  Numeric,2 421.0833 290        "two.sided" "One Sample t-test" "df[i, -1]"
 [9,] 3.292182   11        0.007176877 Numeric,2 421.0833 309        "two.sided" "One Sample t-test" "df[i, -1]"
[10,] 1.471079   11        0.1692895   Numeric,2 421.0833 371        "two.sided" "One Sample t-test" "df[i, -1]"
[11,] -1.75991   11        0.106165    Numeric,2 421.0833 481        "two.sided" "One Sample t-test" "df[i, -1]"
[12,] -3.375404  11        0.006192718 Numeric,2 421.0833 536        "two.sided" "One Sample t-test" "df[i, -1]"

【讨论】：

【解决方案2】：

这是宽数据集挑战的一个很好的例子，其中月份值是单独的列，而不是指标中的值，month，具有与其相邻的相应数值的字段。保持数据较长，包括t.test 在内的大多数跨相关数据的操作都更容易运行。

因此，作为替代方案，考虑重塑，然后为每个 ID 运行 by，在 12 个月 mu 值中迭代调用 t.test：

reshape_df <- reshape(df, varying = names(df)[-1], idvar="id", v.names="value",
                      times = names(df)[-1], timevar="month",
                      new.row.names = 1:1000, direction = "long")

reshape_ref <- reshape(ref, varying = names(ref), v.names="mu", 
                       times = names(ref), timevar="month",
                       new.row.names = 1:1000, direction = "long")

ttest_list <- by(reshape_df, reshape_df$id, function(sub) 
                 do.call(rbind, lapply(reshape_ref$mu, function(x) t.test(sub$value, mu=x))))

输出

ttest_list$`1030879980`

      statistic parameter p.value      conf.int  estimate null.value alternative method              data.name  
 [1,] -18.52462 11        1.214017e-09 Numeric,2 76.08333 507        "two.sided" "One Sample t-test" "sub$value"
 [2,] -18.30968 11        1.375191e-09 Numeric,2 76.08333 502        "two.sided" "One Sample t-test" "sub$value"
 [3,] -15.25747 11        9.526113e-09 Numeric,2 76.08333 431        "two.sided" "One Sample t-test" "sub$value"
 [4,] -15.17149 11        1.01107e-08  Numeric,2 76.08333 429        "two.sided" "One Sample t-test" "sub$value"
 [5,] -16.03127 11        5.649016e-09 Numeric,2 76.08333 449        "two.sided" "One Sample t-test" "sub$value"
 [6,] -12.54917 11        7.334351e-08 Numeric,2 76.08333 368        "two.sided" "One Sample t-test" "sub$value"
 [7,] -14.18274 11        2.05238e-08  Numeric,2 76.08333 406        "two.sided" "One Sample t-test" "sub$value"
 [8,] -9.196035 11        1.696865e-06 Numeric,2 76.08333 290        "two.sided" "One Sample t-test" "sub$value"
 [9,] -10.01282 11        7.298909e-07 Numeric,2 76.08333 309        "two.sided" "One Sample t-test" "sub$value"
[10,] -12.67813 11        6.598342e-08 Numeric,2 76.08333 371        "two.sided" "One Sample t-test" "sub$value"
[11,] -17.40691 11        2.357586e-09 Numeric,2 76.08333 481        "two.sided" "One Sample t-test" "sub$value"
[12,] -19.7713  11        6.045928e-10 Numeric,2 76.08333 536        "two.sided" "One Sample t-test" "sub$value"


ttest_list$`1030879990`

     statistic  parameter p.value     conf.int  estimate null.value alternative method              data.name  
 [1,] -2.523598  11        0.02829384  Numeric,2 421.0833 507        "two.sided" "One Sample t-test" "sub$value"
 [2,] -2.376735  11        0.03671235  Numeric,2 421.0833 502        "two.sided" "One Sample t-test" "sub$value"
 [3,] -0.2912785 11        0.7762574   Numeric,2 421.0833 431        "two.sided" "One Sample t-test" "sub$value"
 [4,] -0.2325333 11        0.8203937   Numeric,2 421.0833 429        "two.sided" "One Sample t-test" "sub$value"
 [5,] -0.8199858 11        0.4296343   Numeric,2 421.0833 449        "two.sided" "One Sample t-test" "sub$value"
 [6,] 1.559197   11        0.1472385   Numeric,2 421.0833 368        "two.sided" "One Sample t-test" "sub$value"
 [7,] 0.4430371  11        0.6663254   Numeric,2 421.0833 406        "two.sided" "One Sample t-test" "sub$value"
 [8,] 3.850262   11        0.00269828  Numeric,2 421.0833 290        "two.sided" "One Sample t-test" "sub$value"
 [9,] 3.292182   11        0.007176877 Numeric,2 421.0833 309        "two.sided" "One Sample t-test" "sub$value"
[10,] 1.471079   11        0.1692895   Numeric,2 421.0833 371        "two.sided" "One Sample t-test" "sub$value"
[11,] -1.75991   11        0.106165    Numeric,2 421.0833 481        "two.sided" "One Sample t-test" "sub$value"
[12,] -3.375404  11        0.006192718 Numeric,2 421.0833 536        "two.sided" "One Sample t-test" "sub$value"

【讨论】：