Personal Loan Project


Predictions Using Machine Learning

In today's competitive banking landscape, identifying potential personal loan customers is a crucial strategy. AllLife Bank, is used here as an example for this Machine Learning project. Let’s assume this bank is a rapidly growing financial institution, seeking my advice using data-driven insights to optimize its loan portfolio.

to breakdown the steps that I’ve taken for this project, follow along…

First, I started with Data Exploration to find any hidden patterns.

As I expected the data showed me several key trends:

  • Income Matters: For instance, customers earning over $100,000 annually were significantly more likely to apply for personal loans.

  • Existing Relationships: Individuals who already had credit cards or savings accounts with AllLife Bank were more receptive to loan offers.

  • Demographics: While age and education played less significant roles, certain demographic segments, such as customers aged 35-45, demonstrated higher interest in personal loans.

>> this is a huge advantage for the bank to consider while improving the loan application. I’d ask about the income and make it a requirement and not so much about the education level, maybe ill make it as an optional question.

The darker shades of green reflects a strong positive relationship. Example, personal loan and income. Customers with higher incomes are more likely to have personal loans.


So, my goal is to develop a Model that acts as a Predictive Tool, follow along…

A decision tree model was carefully crafted to predict personal loan eligibility. This interpretable model was trained on a meticulously prepared dataset, ensuring its accuracy and reliability.

When the model receives an application from a customer inquiring about a personal loan, the file will follow a specific path to get the approval or denial based on the trained model. The blue boxes reflect the approval decisions vs the orange boxes reflect the denial decisions. This is called Machine Learning. cool, is not it?

After Evaluation, the model showed a successful scores!

The model's performance was carefully evaluated, demonstrating impressive accuracy, precision, and recall in identifying potential loan customers.

Key Findings:

  • Income: The primary driver of personal loan interest.

  • Existing Relationships: A significant factor influencing loan consideration. Let's say a customer has been with AllLife Bank for five years and has a checking account, a savings account, and a credit card. This customer is more likely to consider a personal loan from AllLife Bank compared to a customer who has only recently opened an account.

    This is because the existing customer has a longer relationship with the bank, which can build trust and loyalty. Additionally, the bank may have a better understanding of this customer's financial situation, making them more likely to approve a loan application.

My advice:

  • Prioritize High-Income Customers: Focus marketing efforts on individuals earning over $100,000 annually.

  • Leverage Existing Relationships: Offer personal loans to customers who already have credit cards or savings accounts with AllLife Bank.

  • Consider Demographic Factors: Target specific age groups or educational levels based on identified trends.

  • Utilize the Model: Deploy the model to pre-screen loan applications, streamlining the process and improving efficiency.

The predictive model developed in this project would empower the bank to make data-driven decisions, tailor marketing efforts, and of course achieve its business objectives.