摘要统计节省时间
Time Saver on Summary Stats
目标不是交叉制表。我的目标是 table 具有数据集中所有变量的描述性统计信息（在本例中为计数和百分比），这样我就不需要键入 table（变量名称）大约。在处理新数据集时重复 20 次。我计划只使用 xtable + knitr 导出它。不幸的是，循环有问题。任何帮助，将不胜感激。这个问题与那些反复需要提供汇总统计数据的人非常相关，因为这会节省很多时间。很抱歉第三次问这个问题:) 我听说过 table() 函数 - 事实上，我在下面多次使用它 简而言之，这个函数显示变量的名称、它们的级别、计数和百分比，而无需为数据集中的每个变量输入 table() -
ESRD <- rep(c("Y", "N"), each=10)
DIABETES <- rep(c("Y", "N", "Y", "N"), c(5, 5, 5, 5))
BLAH <- rep(c("Y", "N"), each=10)
categoricalvariables <- data.frame(ESRD, DIABETES, BLAH)
descriptives <- function(VARIABLEMATRIX){
desc <- matrix(0, ncol=4, nrow=2*ncol(VARIABLEMATRIX) + ncol(VARIABLEMATRIX))
for (i in 1:ncol(VARIABLEMATRIX)){
matper <- matrix(0, nrow=dim(table(VARIABLEMATRIX[ ,i])), ncol=1)
for (i in 1:dim(table(VARIABLEMATRIX[ ,i]))){
matper[i, ] <- paste(round(prop.table(table(VARIABLEMATRIX[ ,i]))[i]*100, 2), "%")
}
matcount <- matrix(0, nrow=dim(table(VARIABLEMATRIX[ ,i])), ncol=1)
for (i in 1:dim(table(VARIABLEMATRIX[ ,i]))){
matcount[i, ] <- table(VARIABLEMATRIX[ ,i])[i]
}
desc[((3*i)-2), ] <- c(colnames(VARIABLEMATRIX)[i], "", "", "")
desc[((3*i)-1):(3*i), ] <- cbind("", names(table(VARIABLEMATRIX[ ,i])), matcount[ ,1], matper[ ,1])
return(desc)
}
}
descriptives(categoricalvariables)
我得到的输出是（显然有一个错误，但我不确定哪里出了问题）：
[,1] [,2] [,3] [,4]
[1,] "0" "0" "0" "0"
[2,] "0" "0" "0" "0"
[3,] "0" "0" "0" "0"
[4,] "DIABETES" "" "" ""
[5,] "" "N" "10" "50 %"
[6,] "" "Y" "10" "50 %"
[7,] "0" "0" "0" "0"
[8,] "0" "0" "0" "0"
[9,] "0" "0" "0" "0"
预期的输出应该是：
[,1] [,2] [,3] [,4]
[1,] "ESRD" "" "" ""
[2,] "" "N" "10" "50 %"
[3,] "" "Y" "10" "50 %"
[4,] "DIABETES" "" "" ""
[5,] "" "N" "10" "50 %"
[6,] "" "Y" "10" "50 %"
[7,] "BLAH" "" "" ""
[8,] "" "N" "10" "50 %"
[9,] "" "Y" "10" "50 %"
以下是一些使用 tidyverse 函数的选项：
library(tidyverse)
categoricalvariables %>%
gather(Measure, Value) %>%
group_by(Measure, Value) %>%
tally %>%
mutate(Percent=n/sum(n))
Measure Value n Percent
1 BLAH N 10 0.5
2 BLAH Y 10 0.5
3 DIABETES N 10 0.5
4 DIABETES Y 10 0.5
5 ESRD N 10 0.5
6 ESRD Y 10 0.5
categoricalvariables %>%
gather(Measure, Value) %>%
group_by(Measure, Value) %>%
tally %>%
mutate(Percent=n/sum(n)) %>%
gather(Stats, Value2, -Measure, -Value) %>%
unite(Value_Stats, Stats, Value) %>%
spread(Value_Stats, Value2)
Measure n_N n_Y Percent_N Percent_Y
1 BLAH 10 10 0.5 0.5
2 DIABETES 10 10 0.5 0.5
3 ESRD 10 10 0.5 0.5
我保留了数字格式的数据，以备您进行进一步处理。
要为导出的内容设置 table，可能是这样的：
tab = categoricalvariables %>%
gather(Measure, Value) %>%
group_by(Measure, Value) %>%
summarise(Count=n()) %>%
mutate(Percent=paste0(sprintf("%1.1f", Count/sum(Count)*100),"%")) %>%
ungroup %>%
mutate(Measure = ifelse(duplicated(Measure),"", Measure))
Measure Value Count Percent
1 BLAH N 10 50.0%
2 Y 10 50.0%
3 DIABETES N 10 50.0%
4 Y 10 50.0%
5 ESRD N 10 50.0%
6 Y 10 50.0%
现在您可以在 tab 上 运行 xtable。
library(xtable)
print(xtable(tab, align="llcrr"), include.rownames=FALSE)
从 rmarkdown 文档输出到 PDF 时看起来像这样：
如果你有想要汇总的数字列，你可以这样做，例如（使用内置的 iris 数据框）：
iris %>% group_by(Species) %>%
summarise_all(funs(mean, min, max)) %>%
gather(key, value, -Species) %>%
separate(key, c("Measure","Stat"),"_") %>%
spread(Stat, value)
您可能想要进一步重塑它或重新格式化以输出到 table，但它让您了解什么是可能的。
Species Measure max mean min
1 setosa Petal.Length 1.9 1.462 1.0
2 setosa Petal.Width 0.6 0.246 0.1
3 setosa Sepal.Length 5.8 5.006 4.3
4 setosa Sepal.Width 4.4 3.428 2.3
5 versicolor Petal.Length 5.1 4.260 3.0
6 versicolor Petal.Width 1.8 1.326 1.0
7 versicolor Sepal.Length 7.0 5.936 4.9
8 versicolor Sepal.Width 3.4 2.770 2.0
9 virginica Petal.Length 6.9 5.552 4.5
10 virginica Petal.Width 2.5 2.026 1.4
11 virginica Sepal.Length 7.9 6.588 4.9
12 virginica Sepal.Width 3.8 2.974 2.2
descriptr 包中的 oway_tables 函数创建多个单向表。以下是您示例的输出：
> ESRD <- rep(c("Y", "N"), each=10)
> DIABETES <- rep(c("Y", "N", "Y", "N"), c(5, 5, 5, 5))
> BLAH <- rep(c("Y", "N"), each=10)
> categoricalvariables <- data.frame(ESRD, DIABETES, BLAH)
> descriptr::oway_tables(categoricalvariables)
Variable: ESRD
|--------------------------------------------------------------------------|
| Cumulative Cumulative |
| Levels | Frequency | Frequency | Percent | Percent |
|--------------------------------------------------------------------------|
| N | 10 | 10 | 50 | 50 |
|--------------------------------------------------------------------------|
| Y | 10 | 20 | 50 | 100 |
|--------------------------------------------------------------------------|
Variable: DIABETES
|--------------------------------------------------------------------------|
| Cumulative Cumulative |
| Levels | Frequency | Frequency | Percent | Percent |
|--------------------------------------------------------------------------|
| N | 10 | 10 | 50 | 50 |
|--------------------------------------------------------------------------|
| Y | 10 | 20 | 50 | 100 |
|--------------------------------------------------------------------------|
Variable: BLAH
|--------------------------------------------------------------------------|
| Cumulative Cumulative |
| Levels | Frequency | Frequency | Percent | Percent |
|--------------------------------------------------------------------------|
| N | 10 | 10 | 50 | 50 |
|--------------------------------------------------------------------------|
| Y | 10 | 20 | 50 | 100 |
|--------------------------------------------------------------------------|
Link 到函数的文档：oway_tables.
目标不是交叉制表。我的目标是 table 具有数据集中所有变量的描述性统计信息（在本例中为计数和百分比），这样我就不需要键入 table（变量名称）大约。在处理新数据集时重复 20 次。我计划只使用 xtable + knitr 导出它。不幸的是，循环有问题。任何帮助，将不胜感激。这个问题与那些反复需要提供汇总统计数据的人非常相关，因为这会节省很多时间。很抱歉第三次问这个问题:) 我听说过 table() 函数 - 事实上，我在下面多次使用它 简而言之，这个函数显示变量的名称、它们的级别、计数和百分比，而无需为数据集中的每个变量输入 table() -
ESRD <- rep(c("Y", "N"), each=10)
DIABETES <- rep(c("Y", "N", "Y", "N"), c(5, 5, 5, 5))
BLAH <- rep(c("Y", "N"), each=10)
categoricalvariables <- data.frame(ESRD, DIABETES, BLAH)
descriptives <- function(VARIABLEMATRIX){
desc <- matrix(0, ncol=4, nrow=2*ncol(VARIABLEMATRIX) + ncol(VARIABLEMATRIX))
for (i in 1:ncol(VARIABLEMATRIX)){
matper <- matrix(0, nrow=dim(table(VARIABLEMATRIX[ ,i])), ncol=1)
for (i in 1:dim(table(VARIABLEMATRIX[ ,i]))){
matper[i, ] <- paste(round(prop.table(table(VARIABLEMATRIX[ ,i]))[i]*100, 2), "%")
}
matcount <- matrix(0, nrow=dim(table(VARIABLEMATRIX[ ,i])), ncol=1)
for (i in 1:dim(table(VARIABLEMATRIX[ ,i]))){
matcount[i, ] <- table(VARIABLEMATRIX[ ,i])[i]
}
desc[((3*i)-2), ] <- c(colnames(VARIABLEMATRIX)[i], "", "", "")
desc[((3*i)-1):(3*i), ] <- cbind("", names(table(VARIABLEMATRIX[ ,i])), matcount[ ,1], matper[ ,1])
return(desc)
}
}
descriptives(categoricalvariables)
我得到的输出是（显然有一个错误，但我不确定哪里出了问题）：
[,1] [,2] [,3] [,4]
[1,] "0" "0" "0" "0"
[2,] "0" "0" "0" "0"
[3,] "0" "0" "0" "0"
[4,] "DIABETES" "" "" ""
[5,] "" "N" "10" "50 %"
[6,] "" "Y" "10" "50 %"
[7,] "0" "0" "0" "0"
[8,] "0" "0" "0" "0"
[9,] "0" "0" "0" "0"
预期的输出应该是：
[,1] [,2] [,3] [,4]
[1,] "ESRD" "" "" ""
[2,] "" "N" "10" "50 %"
[3,] "" "Y" "10" "50 %"
[4,] "DIABETES" "" "" ""
[5,] "" "N" "10" "50 %"
[6,] "" "Y" "10" "50 %"
[7,] "BLAH" "" "" ""
[8,] "" "N" "10" "50 %"
[9,] "" "Y" "10" "50 %"
以下是一些使用 tidyverse 函数的选项：
library(tidyverse)
categoricalvariables %>%
gather(Measure, Value) %>%
group_by(Measure, Value) %>%
tally %>%
mutate(Percent=n/sum(n))
Measure Value n Percent 1 BLAH N 10 0.5 2 BLAH Y 10 0.5 3 DIABETES N 10 0.5 4 DIABETES Y 10 0.5 5 ESRD N 10 0.5 6 ESRD Y 10 0.5
categoricalvariables %>%
gather(Measure, Value) %>%
group_by(Measure, Value) %>%
tally %>%
mutate(Percent=n/sum(n)) %>%
gather(Stats, Value2, -Measure, -Value) %>%
unite(Value_Stats, Stats, Value) %>%
spread(Value_Stats, Value2)
Measure n_N n_Y Percent_N Percent_Y 1 BLAH 10 10 0.5 0.5 2 DIABETES 10 10 0.5 0.5 3 ESRD 10 10 0.5 0.5
我保留了数字格式的数据，以备您进行进一步处理。
要为导出的内容设置 table，可能是这样的：
tab = categoricalvariables %>%
gather(Measure, Value) %>%
group_by(Measure, Value) %>%
summarise(Count=n()) %>%
mutate(Percent=paste0(sprintf("%1.1f", Count/sum(Count)*100),"%")) %>%
ungroup %>%
mutate(Measure = ifelse(duplicated(Measure),"", Measure))
Measure Value Count Percent 1 BLAH N 10 50.0% 2 Y 10 50.0% 3 DIABETES N 10 50.0% 4 Y 10 50.0% 5 ESRD N 10 50.0% 6 Y 10 50.0%
现在您可以在 tab 上 运行 xtable。
library(xtable)
print(xtable(tab, align="llcrr"), include.rownames=FALSE)
从 rmarkdown 文档输出到 PDF 时看起来像这样：
如果你有想要汇总的数字列，你可以这样做，例如（使用内置的 iris 数据框）：
iris %>% group_by(Species) %>%
summarise_all(funs(mean, min, max)) %>%
gather(key, value, -Species) %>%
separate(key, c("Measure","Stat"),"_") %>%
spread(Stat, value)
您可能想要进一步重塑它或重新格式化以输出到 table，但它让您了解什么是可能的。
Species Measure max mean min 1 setosa Petal.Length 1.9 1.462 1.0 2 setosa Petal.Width 0.6 0.246 0.1 3 setosa Sepal.Length 5.8 5.006 4.3 4 setosa Sepal.Width 4.4 3.428 2.3 5 versicolor Petal.Length 5.1 4.260 3.0 6 versicolor Petal.Width 1.8 1.326 1.0 7 versicolor Sepal.Length 7.0 5.936 4.9 8 versicolor Sepal.Width 3.4 2.770 2.0 9 virginica Petal.Length 6.9 5.552 4.5 10 virginica Petal.Width 2.5 2.026 1.4 11 virginica Sepal.Length 7.9 6.588 4.9 12 virginica Sepal.Width 3.8 2.974 2.2
descriptr 包中的 oway_tables 函数创建多个单向表。以下是您示例的输出：
> ESRD <- rep(c("Y", "N"), each=10)
> DIABETES <- rep(c("Y", "N", "Y", "N"), c(5, 5, 5, 5))
> BLAH <- rep(c("Y", "N"), each=10)
> categoricalvariables <- data.frame(ESRD, DIABETES, BLAH)
> descriptr::oway_tables(categoricalvariables)
Variable: ESRD
|--------------------------------------------------------------------------|
| Cumulative Cumulative |
| Levels | Frequency | Frequency | Percent | Percent |
|--------------------------------------------------------------------------|
| N | 10 | 10 | 50 | 50 |
|--------------------------------------------------------------------------|
| Y | 10 | 20 | 50 | 100 |
|--------------------------------------------------------------------------|
Variable: DIABETES
|--------------------------------------------------------------------------|
| Cumulative Cumulative |
| Levels | Frequency | Frequency | Percent | Percent |
|--------------------------------------------------------------------------|
| N | 10 | 10 | 50 | 50 |
|--------------------------------------------------------------------------|
| Y | 10 | 20 | 50 | 100 |
|--------------------------------------------------------------------------|
Variable: BLAH
|--------------------------------------------------------------------------|
| Cumulative Cumulative |
| Levels | Frequency | Frequency | Percent | Percent |
|--------------------------------------------------------------------------|
| N | 10 | 10 | 50 | 50 |
|--------------------------------------------------------------------------|
| Y | 10 | 20 | 50 | 100 |
|--------------------------------------------------------------------------|
Link 到函数的文档：oway_tables.