5 Types of Data Analysis Problems - Explained with Real Examples

In the world of data, questions come in many forms, but all of them lead to one goal: insights. Whether you're optimizing marketing spend, tracking customer behavior, or forecasting future sales, data analysis provides the tools and mindset to solve different categories of problems.

In this blog, I’ll explain the five types of data analysis problems, show realistic examples using an e-commerce dataset, and explore how each type can be addressed with charts, visualizations, and smart analytical thinking.

1. Descriptive Problems, "What Happened?"

Descriptive analysis focuses on summarizing past events. It helps you understand trends, patterns, and overall performance.

For example, imagine your online perfume store had over 1,000 orders last quarter. That’s a descriptive fact. You get even more informative summaries when you break that down by product category, region, or customer type.

You might discover that: Sales peaked in December due to the holiday season, or "Velvet Oud" was your top seller in the last quarter, and even 40% of your buyers are returning customers.

This is usually presented in charts like bar graphs, pie charts, or KPIs. It sets the foundation for further analysis by laying out the "what".

We might visualize this with a bar chart:

Or present a table like this:

Month	Total Orders	Revenue (₦)
Jan	380	1200000
Feb	450	1500000
Mar	410	1300000

📌Descriptive analytics does not predict or explain why things happened; it only tells us what did.

2. Diagnostic Problems, "Why Did It Happen?"

Once you know what happened, the next natural step is to ask why. This is where diagnostic analysis comes in.

Let’s say your March sales dipped significantly compared to February. Diagnostic analysis would involve digging into:

1. Inventory logs: Was a top-selling item out of stock?

2. Marketing campaigns: Did you spend less on advertising that month?

3. Customer behavior: Was there a drop in website traffic?

You’ll likely use filters, segmentations, and drill-downs to uncover these answers. It often leads to insights like:
“Sales dropped in March because our Facebook ad campaign ended in mid-February, reducing traffic from that channel by 60%.”

This problem type helps with root cause analysis, helping you avoid future mistakes or repeating successes.

🎯 Diagnostic analysis helps identify causes, using tools like filtering, correlations, and drill-downs.

3. Predictive Problems, "What Could Happen Next?"

This type of analysis focuses on the future. Using patterns in historical data, predictive analysis helps forecast likely outcomes.

Say you’re planning stock for July. Based on the trend of rising sales in May and June, you could forecast a 15% increase in demand.

Using techniques like:

1. Trend lines in Excel

2. Time-series models

3. Machine learning regressions

You might predict: Estimated revenue for next month, Expected order volumes, Probability of repeat purchases.

It’s not about certainties; it’s about preparing for what's likely, based on data.

🧠 Predictive analytics often uses regression models or machine learning, but even Excel trendlines can get you started.

4. Prescriptive Problems, "What Should We Do?"

This is where analytics becomes actionable. Prescriptive analysis recommends strategies or actions based on the insights from the other types.

For example, if your diagnostic analysis shows that weekend traffic converts better, and predictive analysis shows rising demand in July, prescriptive analysis might say:

1. Increase ad spend specifically for Friday to Sunday.

2. Run a limited-time "Weekend Exclusive" discount.

3. Ensure your inventory is stocked for peak products.

It often involves what-if analysis, scenario modeling, and simulations.

It answers questions like: “If we increase our budget by 20%, how much more revenue can we generate?”, “Which regions should we focus on to maximize ROI?”.

💡 Prescriptive analytics is essential for optimization and decision-making. It is powerful in operations, inventory management, and pricing strategy.

5. Exploratory Problems, "What Interesting Patterns Can We Find?"

This is the most open-ended type of analysis and often the most exciting. Exploratory Data Analysis (EDA) is about discovery. You're not looking to answer a specific question, you’re trying to uncover hidden patterns.

You might stumble upon insights like:

1. Customers who buy “Musk Royale” also tend to purchase “Cedar Intense” within 30 days.

2. Abuja customers order more often but spend less per order than customers in Lagos.

3. First-time buyers tend to shop more during Instagram promotions than email campaigns.

You’ll use tools like: Pivot tables, Heatmaps, Scatter plots, and box plots

🤯These discoveries often spark new strategies or hypotheses that drive innovation. Exploratory analysis is especially powerful during early stages of research or when entering new markets.

Final Thoughts

Every data analyst needs to become comfortable identifying these types of problems, and more importantly, learning how to switch between them. The magic of data analysis isn’t just about numbers; it’s about asking the right questions at the right time.

As I continue my journey, I’m learning that the true value of data lies in the clarity it brings. Whether you're trying to improve sales, serve customers better, or just understand patterns, data analysis offers a structured, strategic way to get answers.

Let your data do the talking, you just need to ask the right questions.