Fuzzy Matching

[josh1400]

I'm attempting to group duplicate customer names with varying spellings using agrep. I'm new to this so the following might not be ideal, but I'm having trouble just saving the results as a dataset.

This line of code creates a list that seems like it would be useful, but it doesn't save the dataset.

sapply(customer_data$End_Customer, agrep, customer_data$End_Customer)

This line of code doesn't show or save anything, but it does take just as long to run.

fuzzy <- sapply(customer_data$End_Customer, agrep, customer_data$End_Customer)
register("fuzzy")

I've had issues saving datasets in the past, but you helped me resolve them and I don't think I'm making the same mistake. I thought maybe the issue is that a list is created rather than a dataframe so I used this code.

list <- sapply(customer_data$End_Customer, agrep, customer_data$End_Customer)
frame <- data.frame(matrix(unlist(list), nrow=length(list), byrow=TRUE),stringsAsFactors=FALSE)
register("frame")

This saved a dataset, but it didn't have the customer names like I see in the list.

I'll keep troubleshooting, but any advice is appreciated. :)

[josh1400]

I've made this work.

list <- sapply(wa$End_Customer, agrep, wa$End_Customer)
result <- data.frame(stack(list))
register("result")

[josh1400]

This is my first step in a full Fuzzy Matching solution. Any advice for finishing my code will be super helpful!

[vnijs]

The new MSBA cohort arrives on campus this week so I'm swamped. Have you considered posting this to StackOverflow? There are a lot of R folks that could help you there.

[josh1400]

I've gotten this far from StackOverflow. I'll post and see what happens. If you can help when you're more free, this is something I'll be using for a lot of datasets at Micron.

Speaking of MSBA, I'd like to sit in on one of your classes. I haven't seen any possibilities yet from the emails Rady sends out. Do you plan on having any classes this upcoming school year where you dive more into R to add onto what I've learned since graduating?

[vnijs]

Reach out to me by email and we can discuss.