bank

Predicting the drivers for bad loans for a leading German Bank

Objective: One of the leading banks in Germany would like to solve the following business problem:

1. The drivers for the bad loans and predict the bad loans given the drivers.

Data Dictionary: Number of variables in the dataset: 21

1. Badloan: 0: Good Loan 1: Bad Loan
2. Account Balance: Status of existing checking account 1: Less than 0 USD 2: Between 0 and 200 USD 3: Greater than 200 USD 4: No checking account
3. Duration of credit (months)
4. Payment status of previous credit: Credit History 0: No credits taken/All credits paid back fully 1:All credits at bank paid back duly 2: existing credits paid back duly till now 3:delay in paying off in the past 4: critical account/ other credits existing(not at this bank)
5. Purpose 0: Car (new) 1: Car (used) 2: furniture/equipment 3: radio/television 4: domestic appliances 5: repairs 6: education 7: vacation 8: retraining 9: business 10: others
6. Credit Amount: Loan Amount
7. Value Savings/Stocks 1: Less than 100 DM 2: Between 100 and 500 DM 3: Between 500 and 1000 DM 4: Greater than or equal to 1000 DM
8. Length of current employment 1: Unemployed 2: Less than one year 3: Between 1 and 4 years 4: Between 4 and 7 years 5: Greater than or equal to 7 years
9. Instalment per cent
10. Sex & Marital status 1: Male: Divorced/Separated 2: Female: Divorced/Separated/Married 3: Male: Single 4: Male: Married/widowed 5: female: single
11. Guarantors 1: None 2: Co-applicant 3: guarantor
12. Duration of current address
13. Most valuable available asset 1: Real estate 2: if not 1: building society savings account/life insurance 3: if not1 or 2: Car or other 4: unknown/no property
14. Age(years)
15. Concurrent credits 1: bank 2: stores 3: none
16. Type of apartment 1: rent 2: own 3: for free
17. No. of credits at this bank
18. Occupation 1: unemployed/unskilled – non-resident 2: unskilled – resident 3: skilled employee / official 4: management/ self – employed / highly qualified employee / officer
19. Number of dependents
20. Telephone 1: none 2: yes
21. Foreign Worker 1: yes 2: no

Methodology:

1. Data cleaning and visualization
2. Machine learning model building
3. Feature Selection
4. Machine learning model building with transformations
5. Model validation and comparison
6. Results and conclusion

Essential Packages:

1. caret
2. ggplot2
3. caretEnsemble
4. readr