Groupby node use similar to R Groupby()

So I have R code that is a table of 12 variables.

The code groups the table by a number of variables:

groupby( daysncinstall, mediasource, cohortday, weekinstal, name, app_id)

The resulting table is:

I then summarize values with the following code:

summarise(totdayearn=sum(value), users=sum(users), peruser=(totday_earn/users))

The resulting table is:

So I’m trying to replicate this in Knime. When I try in Knime, if I do all the grouping in one Groupby node:

I end up with this before the join to the original data:

But after the join, I end up with 25 million rows rather than the 15,364 rows from the R code.

The join configuration is:

Is my groupby with Knime nodes incorrect or is the joiner setup incorrect? Do I need to do smaller successive groupbys joining each with the original table?