Page 48 - Banking Finance October 2025

A R T IC L E
ARTICLE
3. Predictive Analytics - This type uses historical data to without consent or exploiting sensitive information, can
predict future outcomes. For example, banks use damage reputation and violate ethical standards.
predictive analytics to forecast loan defaults by Analysing purchasing behaviour to aggressively target
analysing past borrower behaviour and economic vulnerable customers with high-interest loans could be
indicators. This helps in proactive risk management. considered exploitative.
4. Prescriptive Analytics - This type suggests actions to 6. Cherry-Picking Data
achieve desired outcomes. For instance, a bank might Selectively using data that supports a desired outcome
use prescriptive analytics to recommend personalized or ignoring inconvenient data can lead to biased
financial products to customers based on their conclusions and flawed decision-making. Analysts may
transaction history and financial goals. unknowingly favour data or interpretations that confirm
their pre-existing beliefs or hypotheses, leading to
Key considerations in data analytics skewed conclusions.
While data analytics offers substantial benefits, its 7. Misinterpreting or Miscommunicating Results
effectiveness could be impacted by Poor interpretation or communication of data analytics
1. Poor Data Quality results can lead to incorrect conclusions, causing
Inaccurate, incomplete, or outdated data can lead to decision-makers to act on faulty insights. Presenting
flawed analysis and unreliable insights. "Garbage in, average values without considering outliers can mask
garbage out" is a critical issue in analytics. important insights, such as extreme customer
satisfaction or dissatisfaction.
2. Data-Driven Culture and investment in Robust
Infrastructure 8. Failing to Iterate and Update Models
Securing commitment from senior management to Treating a model as "done" after its initial deployment
prioritize data analytics initiatives, inculcating data- can lead to outdated insights, especially in dynamic
driven behaviours among employees and implementing environments. A predictive model for customer churn
scalable data storage and processing systems to built using last year's data might not perform well as
accommodate growing data volumes and evolving customer preferences or market conditions change.
analytical needs are fundamental for proper data
analytics. The Complete Cycle of Data Analytics: A Detailed Example
To illustrate the complete data analytics lifecycle, let's
3. Correlation vs. Causation
explore a detailed example related to the banking sector:
Misinterpreting correlation as causation is a common
Predicting Loan Defaults to Enhance Credit Risk
mistake in data analytics. Just because two variables
Management.
move together does not mean one causes the other. A
bank might observe that customers with certain
spending habits tend to default on loans, but that Scenario
doesn't mean those habits cause the default. A bank aims to reduce loan defaults by implementing a data-
driven approach to assess the creditworthiness of loan
4. Ignoring the Context or Domain Expertise
Data analytics without domain knowledge can lead to applicants. By predicting the likelihood of default, the bank
can make informed lending decisions, mitigate risks, and
incorrect assumptions and conclusions. Data might tell
improve financial performance.
a partial story, but without the right context,
interpretations may be misleading. A spike in customer
churn may be due to external economic factors, but Step-by-Step Implementation
without domain knowledge, the analysis might 1. Data Collection
incorrectly attribute it to internal service issues. Internal Data Sources:
5. Overlooking Data Privacy and Security Customer Demographics: Age, gender, marital
Mishandling sensitive data can lead to privacy breaches status, education, occupation.
and legal issues. Analytics that involve ethically Financial Information: Income, existing debts,
questionable practices, such as using customer data credit history, loan amount, repayment period.

BANKING FINANCE | OCTOBER | 2025 | 43