根据符合标准的观测数量在R中对数据集进行子集。 [R]

Question

I have a dataset that looks like this:

Employee    Month       CSAT

ABROWN      February    4

ABROWN      January     5

ABROWN      March       3

ABROWN      March       5

JSMITH      February    5

JSMITH      January     3

JSMITH      February    5

JSMITH      March       5

JSMITH      February    5

JSMITH      January     4

Except of course much larger. I'm trying to run analysis on Employee by month, but I don't want to include employees for whom there aren't enough observations in a certain month.

For instance, lets say in this case, I only want to keep observation where an Employee has at least two CSAT scores in the same month. In this case we would filter out observations 1,2, and 8.

I've messed with this for too long. And am at a loss.

Answer 1

We can do this with data.table . Convert the 'data.frame' to 'data.table' ( setDT(df1) ), grouped by 'Employee', 'Month', if the number of observations ( .N ) is greater than 1, Subset the Data.table ( .SD )

library(data.table)
setDT(df1)[, if(.N >1) .SD,  by = .(Employee, Month)]
#   Employee    Month CSAT
#1:   ABROWN    March    3
#2:   ABROWN    March    5
#3:   JSMITH February    5
#4:   JSMITH February    5
#5:   JSMITH February    5
#6:   JSMITH  January    3
#7:   JSMITH  January    4

---

Or using dplyr with similar logic in filter after grouping by 'Employee', 'Month'

library(dplyr)
df1 %>%
   group_by(Employee, Month) %>%
   filter(n() >1)

---

Or using base R with ave to create a logical index filter the rows of 'df1'.

df1[with(df1, ave(CSAT, Employee, Month, FUN=length)>1),]