我有一個包含多個變數的資料框，每個變數在兩個不同的時間點用多個專案測量。我想在包含字串相同部分的列組中洗掉所有帶有 NA 條目的行。其中一些組包含多列（例如，grep("learn")，有些只有一列（例如，T1_age。這是我的原始資料框（其中的一部分）：
data <- data.frame(
T1_age = c(39, 30, 20, 48, 27, 55, 37, 50, 50, 37),
T1_sex = c(2, 1, 1, 2, 2, 1, 1, 2, 1, 1),
T2_learn1 = c(2, NA, 3, 4, 1, NA, NA, 2, 4, 4),
T2_learn2 = c(1, NA, 4, 4, 1, NA, NA, 2, 4, 4),
T2_learn3 = c(2, NA, 4, 4, 1, NA, NA, 3, 4, 4),
T2_learn4 = c(2, NA, 2, 5, 5, NA, NA, 5, 5, 5),
T2_learn5 = c(4, NA, 3, 4, 3, NA, NA, 3, 4, 3),
T2_aut1 = c(NA, NA, 4, 4, 4, NA, NA, 3, 5, 4),
T2_aut2 = c(NA, NA, 4, 4, 4, NA, NA, 3, 5, 5),
T2_aut3 = c(NA, NA, 4, 4, 3, NA, NA, 3, 5, 5),
T2_ssup1 = c(1, NA, 4, 5, 4, NA, NA, 2, 4, 3),
T2_ssup2 = c(3, NA, 4, 5, 5, NA, NA, 3, 4, 4),
T2_ssup3 = c(4, NA, 4, 5, 5, NA, NA, 4, 4, 4),
T2_ssup4 = c(2, NA, 3, 5, 5, NA, NA, 3, 4, 4),
T3_learn1 = c(3, NA, NA, 4, 4, NA, NA, 3, 3, 4),
T3_learn2 = c(1, NA, NA, 4, 3, NA, NA, 3, 3, 4),
T3_learn3 = c(3, NA, NA, 4, 4, NA, NA, 3, 3, 5),
T3_learn4 = c(4, NA, NA, 5, 4, NA, NA, 4, 5, 5),
T3_learn5 = c(4, NA, NA, 3, 4, NA, NA, 3, 3, 4),
T3_aut1 = c(NA, NA, NA, 4, 4, NA, NA, 3, 5, 5),
T3_aut2 = c(NA, NA, NA, 3, 4, NA, NA, 3, 5, 5),
T3_aut3 = c(NA, NA, NA, 3, 2, NA, NA, 3, 5, 5),
T3_ssup1 = c(3, NA, NA, 5, 4, NA, NA, 2, 4, 1),
T3_ssup2 = c(3, NA, NA, 5, 5, NA, NA, 4, 5, 5),
T3_ssup3 = c(4, NA, NA, 5, 5, NA, NA, 4, 5, 3),
T3_ssup4 = c(3, NA, NA, 5, 5, NA, NA, 4, 5, 4)
)
現在我已經找到了一個非常糟糕的解決方案，我相信可以改進。所以這段代碼基本上做了我想要的：
library(dplyr)
library(tidyr)
data <- data %>% filter(rowSums(is.na(.[ , grep("learn", colnames(.))])) != ncol(.[ , grep("learn", colnames(.))]))
data <- data %>% filter(rowSums(is.na(.[ , grep("aut", colnames(.))])) != ncol(.[ , grep("aut", colnames(.))]))
data <- data %>% filter(rowSums(is.na(.[ , grep("ssup", colnames(.))])) != ncol(.[ , grep("ssup", colnames(.))]))
data <- data %>% drop_na(T1_age)
data <- data %>% drop_na(T1_sex)
所以新的資料框（以及我想要實作的）看起來像這樣：
data2 <- data.frame(
T1_age = c(20, 48, 27, 50, 50, 37),
T1_sex = c(1, 2, 2, 2, 1, 1),
T2_learn1 = c(3, 4, 1, 2, 4, 4),
T2_learn2 = c(4, 4, 1, 2, 4, 4),
T2_learn3 = c(4, 4, 1, 3, 4, 4),
T2_learn4 = c(2, 5, 5, 5, 5, 5),
T2_learn5 = c(3, 4, 3, 3, 4, 3),
T2_aut1 = c(4, 4, 4, 3, 5, 4),
T2_aut2 = c(4, 4, 4, 3, 5, 5),
T2_aut3 = c(4, 4, 3, 3, 5, 5),
T2_ssup1 = c(4, 5, 4, 2, 4, 3),
T2_ssup2 = c(4, 5, 5, 3, 4, 4),
T2_ssup3 = c(4, 5, 5, 4, 4, 4),
T2_ssup4 = c(3, 5, 5, 3, 4, 4),
T3_learn1 = c(NA, 4, 4, 3, 3, 4),
T3_learn2 = c(NA, 4, 3, 3, 3, 4),
T3_learn3 = c(NA, 4, 4, 3, 3, 5),
T3_learn4 = c(NA, 5, 4, 4, 5, 5),
T3_learn5 = c(NA, 3, 4, 3, 3, 4),
T3_aut1 = c(NA, 4, 4, 3, 5, 5),
T3_aut2 = c(NA, 3, 4, 3, 5, 5),
T3_aut3 = c(NA, 3, 2, 3, 5, 5),
T3_ssup1 = c(NA, 5, 4, 2, 4, 1),
T3_ssup2 = c(NA, 5, 5, 4, 5, 5),
T3_ssup3 = c(NA, 5, 5, 4, 5, 3),
T3_ssup4 = c(NA, 5, 5, 4, 5, 4)
)
你能幫我改進一下嗎？謝謝！！！
uj5u.com熱心網友回復：
您可以grep在 an 中迭代sapply并檢查rowSums切片中的 s 是否達到它們的列數。
V <- c('learn', 'aut', 'ssup')
res <- data[!rowSums(sapply(V, \(v) {
X <- data[grep(v, names(data))]
rowSums(is.na(X)) == dim(X)[2]
})), ]
stopifnot(all.equal(res, data2, check.attributes=FALSE))
或者可能只是檢查NA“熱”列中's的總和是否達到列數（沒有人口統計資料）就足夠了。
res1 <- data[rowSums(is.na(data[grep(paste(V, collapse='|'), names(data))])) !=
dim(data[-(1:2)])[2], ]
stopifnot(all.equal(res1, data2, check.attributes=FALSE))
data2是您在 OP 中提供的結果資料框。dim(data)[2]給出與 相同ncol(data)。
筆記： R version 4.1.2 (2021-11-01)
uj5u.com熱心網友回復：
下面的代碼應該可以正常作業，因為我已經通過將隨機元素更改為NA. 我所做的是將原始資料子集為 T2 和 T3 資料（因為各自的列號相同），然后通過is.na().
data_T2 <- data %>% subset(select = c(3:14))
data_T3 <- data %>% subset(select = c(15:26))
data[!(is.na(data_T2) & is.na(data_T3))[,1],]
轉載請註明出處，本文鏈接：https://www.uj5u.com/qita/391185.html
標籤：r
下一篇：如何讀取.xpt檔案？