Midterm2: Clustering

Dataset: https://www.kaggle.com/datasets/youssefismail20/credit-and-smes-synth-data-credit-analysis/data

Goals:

Using the synthetic credit card consumers dataset, determine an optimal clustering.

Important Links

• Data Source This is included as a reference, please don’t go to it
• Data to use

Submission

• Part 1:
  • PDF explaining what you tried, and why you are convinced your solution is the best
    • This should include justifications of the type of clustering, the number of clusters, interpretations of the clusters, and how someone could use the cluster information.
  • It should include all models you submitted to leaderboard
• Part 2:
  • Code You used for cleaning the data and fitting the model
  • This should be code and runnable without me downloading anything else, no local references
• Part 3:
  • predicted.csv
    • includes customer ID and cluster assignment only
  • There will be a leaderboard. I am not telling you how the leaderboard is scored
  • You may submit more than once
    • as with Midterm1 if you submit multiple times, you should change your pdf to include what you changed and why

Example

library(cluster)
library(factoextra)
library(tidyverse)

cons_data<-read.csv("https://rebelskyw.cs.grinnell.edu/wp-content/uploads/2025/03/synthetic_credit_consumers.csv")

row.names(cons_data)<-cons_data[,1]
k2 <- kmeans(cons_data[3:7], centers = 2, nstart = 1)
out_data<-data.frame(k2$cluster)
out_data$id<-row.names(out_data)
write_csv(out_data,"predicted.csv")
head(read_csv("predicted.csv")) # your dataset should look like this, make sure you don't save the row id

## Rows: 1000 Columns: 2
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): id
## dbl (1): k2.cluster
## 
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

## # A tibble: 6 × 2
##   k2.cluster id                                  
##        <dbl> <chr>                               
## 1          1 e11fa949-37eb-4999-bdbd-5d3c401b80fa
## 2          1 416c9fe3-d43b-45a5-978e-aa8a4311b5c9
## 3          1 b1b99342-704e-449f-90f7-2c5287518e1f
## 4          1 67a301f6-c1ad-49be-92f6-36e6201c82ac
## 5          2 a90ff7c6-c1ac-4fe5-896f-dfb1685b7a3b
## 6          2 9b2239f3-8556-4c7a-acf8-03ab4f82540a