IS6052 ‐ Descriptive and Predictive Analytics
2022‐2023 Individual CA Project
Due Date: Thursday December 8th
Submit your project report as a single pdf file on Canvas
Loan Appraisal for FNB Bank
You are a credit analyst working for FNB Bank. Your responsibilities include analysing the loan
applications and making recommendations to management, based upon your findings, to help them
make data‐driven decisions on lending.
“Loan.csv” file includes data on 40,000 FNB customers that were granted a loan in the past and the
espective outcome, i.e. whether they were identified as “write‐offs” or “not write‐offs”.
Using this data, analyse the new loan applications whose data are provided in “NewApplications.csv”
and try to predict whether each new applicant will repay the requested loan if it is approved.
Data available:
1. Gender M=male; F=female
2. Age an integer parameter
3. marital_status widowed; ma
ied; single; divorced
4. education basic; highsch; univ; postgrad
5. nb_depend_child number of dependent children (0,1,2,3)
6. employ_status employment status (full_time; part_time; unemployed; self_employ; retired)
7. yrs_cu
ent_job years at the cu
ent employment
8. yrs_employed total number of years employed so far
9. net_income an integer parameter
10. spouse_work yes; no
11. spouse_income if the spouse works, what is his/her income?
12. residential_status home owner (owner); tenant; home owner with a mortgage (owner_morg);
living with parents (w_parents)
13. yrs_cu
ent_address years at the cu
ent address
14. loan_amount an integer parameter
15. loan_purpose debt consolidation (debt_consol); wedding; home improvement (home_improv);
vehicle; holidays; other
16. loan_length the duration of the loan
17. collateral yes; no
18. writeoff yes; no
SAT-Dell2019
Cross-Out
SAT-Dell2019
Typewritten Text
Monday December 19th
Your report should contain:
1. an investigation of the data and a summary of your descriptive analyses; (18 pts)
2. a discussion on the pros and cons of the prediction methods that can be used to address FNB's loan
appraisal problem; (6 pts)
3. a
ief description (and the assumptions made, if any) of how selected prediction methods are
applied; (7 pts)
4. R codes developed; (12 pts)
5. an evaluation of the results obtained by each prediction method tried on the data; (20 pts)
6. a comparative analysis of the results; (20 pts)
7. your final recommendation to FNB on which customers should be granted loan; (10 pts)
8. a discussion on any additional data that you think would be useful, if collected, to make better
predictions in the future. (7 pts)