summarize categorical data based on grouping - r

Question

I have a data frame in the following form
Id <- c(101,102,103,101,103,103,102,101,103,102)
Service <- c('A','B','A','C','A','A','B','C','A','B')
Type <- c('C','I','C','I','C','C','C','I','I','C')
Channel <- c('ATM1','ATM2','ATM1','Teller','Teller','ATM2','ATM1','ATM1','ATM2','Teller')
amount <- c(11,34,56,37,65,83,26,94,34,55)
df <- data.frame(Id,Service,Channel,Type,amount)
df in tabular formate
Id Service Channel Type amount
101 A ATM1 C 11
102 B ATM2 I 34
103 A ATM1 C 56
101 C Teller I 37
103 A Teller C 65
103 A ATM2 C 83
102 B ATM1 C 26
101 C ATM1 I 94
103 A ATM2 I 34
102 B Teller C 55
I am able to summarize my data using amount column as df %>% group_by(Id) %>% summarise(total = sum(amount)) %>% as.data.frame
Id total
101 142
102 115
103 238
How can I summarize data in a similar way using categorical columns (Service/Type/Channel) and group_by(Id)? I know we can use table() here, but I am trying to create a data frame, which I can use it for further analysis, such as clustering.

Answer 1

One way to restructure the categorical variables in a manner that they can be summarized by Id is to create dummy coded variables, where 1 means presence, 0 means absence. Then, aggregate results in counts of each category (i.e. number of times ATM 1 used) by Id.
We use the dummies package to create dummy coded variables.
Id <- c(101,102,103,101,103,103,102,101,103,102)
Service <- c('A','B','A','C','A','A','B','C','A','B')
Type <- c('C','I','C','I','C','C','C','I','I','C')
Channel <- c('ATM1','ATM2','ATM1','Teller','Teller','ATM2','ATM1','ATM1','ATM2','Teller')
amount <- c(11,34,56,37,65,83,26,94,34,55)
df <- data.frame(Id,Service,Channel,Type,amount)
library(dummies)
df <- dummy.data.frame(df,names=c("Service","Type","Channel"))
aggregate(. ~ Id,data=df,"sum")
...and the output:
> aggregate(. ~ Id,data=df,"sum")
Id ServiceA ServiceB ServiceC ChannelATM1 ChannelATM2 ChannelTeller TypeC
1 101 1 0 2 2 0 1 1
2 102 0 3 0 1 1 1 2
3 103 4 0 0 1 2 1 3
TypeI amount
1 2 142
2 1 115
3 1 238
>
We interpret the results as follows.
Id 101 used Service A once, Service C twice, ATM1 once, a Teller once, Type I once, and Type C twice for a total amount of 142.