0% found this document useful (0 votes)
11 views17 pages

Problem Formulation in Data Science

The document outlines the process of problem formulation in data science, emphasizing the importance of understanding domain problems, formulating clear problem statements, and developing testable hypotheses. It provides guidelines for creating SMART problem statements and describes a structured approach to hypothesis generation and testing. Key takeaways include the necessity of domain understanding, the iterative nature of the process, and the significance of asking the right questions.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
11 views17 pages

Problem Formulation in Data Science

The document outlines the process of problem formulation in data science, emphasizing the importance of understanding domain problems, formulating clear problem statements, and developing testable hypotheses. It provides guidelines for creating SMART problem statements and describes a structured approach to hypothesis generation and testing. Key takeaways include the necessity of domain understanding, the iterative nature of the process, and the significance of asking the right questions.
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Problem Formulation in Data Science

Understanding Domain Problems, Problem Statements, and


Hypotheses

Department of Statistics
University of Lagos

First Semester 2025


Last updated: December 10, 2025

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 1 / 17
Contents

1 Understanding the Domain Problem

2 Problem Statement Formulation

3 Hypothesis Formulation

4 The Integrated Process

5 Tools and Templates

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 2 / 17
The Foundation of Any Data Science Project

Key Principle
No amount of sophisticated analytics can compensate for a poorly
understood problem.

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 3 / 17
What is a Domain Problem?

A real-world challenge within a specific field that:


Has measurable impact
Can be addressed through data-driven insights
Exists within constraints (technical, ethical, regulatory, budgetary)

Examples:
Healthcare: Reduce patient readmission rates by 15% within 12
months
Retail: Increase customer lifetime value by identifying high-value
segments
Manufacturing: Predict equipment failure to reduce downtime

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 4 / 17
The Critical First Step: Problem Understanding

Question Purpose
Who Identify stakeholders
What Define problem clearly
Why Quantify impact
Where Determine context
When Identify temporal patterns
How Understand current solutions

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 5 / 17
Characteristics of a Well-Formulated Problem Statement

Follows the SMART framework:


Specific: Clear and unambiguous
Measurable: Quantifiable success criteria
Achievable: Realistic given constraints
Relevant: Aligns with business objectives
Time-bound: Includes timeframe considerations

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 6 / 17
Problem Statement Template

Template
In the context of [DOMAIN], we are facing [SPECIFIC PROBLEM] which
affects [STAKEHOLDERS] resulting in [QUANTIFIED IMPACT]. A
successful solution would [DESIRED OUTCOME] measured by [SUCCESS
METRICS] within [TIMEFRAME].

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 7 / 17
Example Transformations

Poor Formulation Improved Formulation


We need to improve sales Q3 sales in Northeast decreased
by 18% due to churn in 25–35 age
group
Make our website better Mobile users have 65% higher
bounce rate and 40% shorter ses-
sions

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 8 / 17
Types of Hypotheses

1 Research Hypotheses:
Hr : “Customer satisfaction affects retention rates”
2 Statistical Hypotheses:
H0 : “No correlation between satisfaction and retention”
H1 : “Higher satisfaction correlates with higher retention”
3 Machine Learning Hypotheses:
Hml : “Feature X improves prediction accuracy by at least 5%”

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 9 / 17
Hypothesis Development Framework

Observation → Question → Literature Review → Hypothesis


Formulation

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 10 / 17
Hypothesis Prioritization: ICE Scoring

Impact: Expected improvement over baseline


Confidence: Evidence strength
Ease: Feasibility and cost of testing

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 11 / 17
Data Science Process Flow

1 Domain Understanding
2 Problem Statement Formulation
3 Hypothesis Generation
4 Data Requirements
5 Experimental Design
6 Execution & Validation

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 12 / 17
Case Study: E-commerce Platform

Problem Statement
“Our e-commerce platform has a conversion rate 40% below industry
average, especially for mobile users. We need to identify and address the
top 3 barriers to increase mobile conversion to at least 2.0% within 6
months.”
Hypotheses:
1 Mobile checkout has too many steps
2 Slow page load times reduce conversion
3 Difficulty applying discount codes

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 13 / 17
Hypothesis Testing Matrix

Hypothesis Test Design Data Required Success Criteria


Mobile checkout steps A/B Test User logs, funnel data p < 0.05, lift > 5%
Page load times Correlation analysis Performance logs r > 0.3
Discount codes Usability study Survey data, recordings 90% success rate

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 14 / 17
Summary: Key Takeaways

1 Domain understanding comes first


2 Problem statements should be SMART
3 Hypotheses must be testable and falsifiable
4 Iterate and refine continuously
5 Document assumptions and decisions

“The most important step in data science isn’t building mod-


els—it’s asking the right questions.”

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 15 / 17
References

CRISP-DM: Cross-Industry Standard Process for Data Mining


OSEMN Framework: Obtain, Scrub, Explore, Model, Interpret

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 16 / 17
Thank You

First Semester 2025 Last updated: Decemb


Department of Statistics University of Lagos Problem Formulation in Data Science 17 / 17

You might also like