Calculating percentages of a factor variable with dplyr
I am trying to calculate percentages/counts of each level of a factor variable in a data frame within dplyr, kind of like using table, and while I can do this manually, this becomes tedious if I have many factor variables or the factor variable has many levels.
Example:
set.seed(100)
data <- data.frame(groupbyvar = LETTERS[1:4],
var1 = letters[1:4],
var2 = as.factor(sample(1:4,12,TRUE)))
data %>% group_by(groupbyvar) %>% summarise(var1_a = mean(var1 == 'a', na.rm=TRUE),
var1_b = mean(var1 == 'b', na.rm=TRUE),
var1_c = mean(var1 == 'c', na.rm=TRUE),
var1_d = mean(var1 == 'd', na.rm=TRUE),
var1_1 = mean(var2 == 1, na.rm=TRUE),
var1_2 = mean(var2 == 2, na.rm=TRUE),
var1_3 = mean(var2 == 3, na.rm=TRUE),
var1_4 = mean(var2 == 4, na.rm=TRUE))
I thought about using table, but this doesn't generate output that dplyr can understand. Also, I thought about using something like model.matrix to generate indicators on the factor variables before passing in the dataframe, but this increases memory footprint unnecessarily (esp for a large data set). Is there some easy way to automate this?
The result should be a new dataframe with percentages/counts:
groupbyvar var1_a var1_b var1_c var1_d var1_1 var1_2 var1_3 var1_4
1 A 1 0 0 0 0.0000000 0.6666667 0.3333333 0.0000000
2 B 0 1 0 0 0.3333333 0.6666667 0.0000000 0.0000000
3 C 0 0 1 0 0.0000000 0.0000000 0.6666667 0.3333333
4 D 0 0 0 1 0.3333333 0.3333333 0.0000000 0.3333333
I want it to automate the suffix on each column name, similar to what model.matrix does with factor variables.
It's definitely overly complicated but I do think that tables will do what you probably want.
Your data
set.seed(100)
data <- data.frame(groupbyvar = LETTERS[1:4],
var1 = letters[1:4],
var2 = as.factor(sample(1:4,12,TRUE)))
Then we put it into tabular .
mytab<-tabular( (Factor(groupbyvar, "Group") + 1)*(
(ColPct=Percent("col")))
~ (Factor(var1, "var1")
+ Factor(var2, "var2") +
1)
*Format(digits=1), data=data )
Which gives me myTab
var1 var2
Group a b c d 1 2 3 4 All
A ColPct 100 0 0 0 0 40 33 0 25
B ColPct 0 100 0 0 50 40 0 0 25
C ColPct 0 0 100 0 0 0 67 50 25
D ColPct 0 0 0 100 50 20 0 50 25
All ColPct 100 100 100 100 100 100 100 100 100
And then extract the data from myTab (eg class(table(myTab)) is table. It is really a pain though.
下一篇: 用dplyr计算因子变量的百分比