Hasan Ramadan

📊 Statistics | Data & Business Analytics
🎓 Teaching Statistics • Math • Data Analytics • Business
💡 Clear insights & real-world applications

[Hasan Ramadan @HasannRamadan]

One easy example would be declining sales. It’s easy to jump straight into analyzing sales trends and building dashboards, but without understanding the business, you might miss the real questions, is it a pricing issue, lower demand, poor distribution, seasonality, or losing customers to competitors?
The same data can tell very different stories depending on the business context.

[Hasan Ramadan @HasannRamadan]

5 friends compare monthly salaries.

The "average" says they each earn $3,200.

But 4 of the 5 earn less than $1,600.

The math isn't wrong. The tool is. Here's how this happens and how to catch it ↓

The salaries:
$1,000 · $1,200 · $1,300 · $1,500 · $11,000
Total $16,000 ÷ 5 = $3,200. That's the mean, and it's calculated correctly. So why does it describe nobody in the group?

The mean uses the size of every value, so one big number pulls it toward itself. Take out the $11,000 and the mean of the other 4 is $1,250. One person moved the "average" by $1,950.

Now the median: sort the values and take the middle one.
$1,000 · $1,200 · [$1,300] · $1,500 · $11,000

Median = $1,300. Change $11,000 to $1,000,000 and it's still $1,300. Outliers barely move it.
When to use which:
→ Median: skewed data with a few extreme values (salaries, house prices, delivery times)
→ Mean: roughly symmetric data with no big outliers (heights, measurement errors)

A 5-second check for exams and real data:
Mean much bigger than median → a few very large values (right skew)
Mean much smaller than median → a few very small values (left skew)
Mean ≈ median → roughly symmetric

Your turn. Exam scores: 45, 70, 72, 75, 78
Should you report the mean or the median? Which is bigger?

[Hasan Ramadan @HasannRamadan]

@TechByTaraa The easiest is definitely Python but to be honest I would recommend learning Java first.

[Hasan Ramadan @HasannRamadan]

@Tech_p001 I learned Java back in the days and to be honest I'm glad I did because every other language became easier afterwards. Java provide a solid foundation to learn OOP simply. So personally my experience in Java first made me a lot better in every coding language then.

[Hasan Ramadan @HasannRamadan]

@ekemini58110 Programming needs logical thinking close to what mathematics gives you. Also, you use some in advanced levels but to be honest its fine and nothing scary.

[Hasan Ramadan @HasannRamadan]

@JA_Olaoye You are right because sometimes even if we cover a lot of KPIs to solve a certain question or problem after a while if we search deeply, we might find another way to read the data to do new things to help the business.

[Hasan Ramadan @HasannRamadan]

@DataWithObinna Any problem you are trying to solve with data starts with the right questions and KPIs.........
Mainly having a business sense nowadays is really important as a data analyst more than making eyecatching charts and python agents.

[Hasan Ramadan @HasannRamadan]

@kapilansh_twt AI has always been the foremen and the helper of data scientists and analysts but not replacing their roles. As a data analyst AI increased my productivity by 30-40%

[Hasan Ramadan @HasannRamadan]

@ezekiel_aleke It's weird seeing people posting only charts and not posting a problem they solved or a question they answered ...........
I know charts in general are ways we express and answer the topic we are solving or answering but being cool at making charts doesn't solve problems alone.

[Hasan Ramadan @HasannRamadan]

A dashboard with 20 KPIs can tell you less than a dashboard with 5.
For a small business, I’d start with:
• Revenue
• Gross margin
• Average order value
• Best/worst-performing products
• Sales vs previous period
Then ask one question:
“What decision will this number help us make?”
If a KPI doesn’t help you understand performance or make a decision, it may not need to be there.
Business analytics is not about having more charts.
It’s about seeing what matters.

[Hasan Ramadan @HasannRamadan]

@cheftee_lead Very True

[Hasan Ramadan @HasannRamadan]

@cheftee_lead Before even looking at the dataset or request it
you should ask yourself what problem you're trying to solve or what goals you are seeking and from there you start building questions, KPIs, ......

[Hasan Ramadan @HasannRamadan]

@komal_uk01 Did we stop learning mathematics after the calculator came?

[Hasan Ramadan @HasannRamadan]

Statistics isn't just about calculating an average.
Imagine two products both sell an average of 100 units per week.
Product A: 98, 101, 99, 102
Product B: 40, 160, 50, 150
Same average.
Completely different business situation.
That's why variation matters.
Good analysis asks more than:
“What's the average?”
It also asks:
“How consistent is it?”

[Hasan Ramadan @HasannRamadan]

Statistical significance and practical importance are different.
A large sample can make a tiny effect look “significant.”
Report the size of the effect, not only the p-value.

[Hasan Ramadan @HasannRamadan]

@hugzlino It gets easier when you do one example slowly instead of rereading the notes.

[Hasan Ramadan @HasannRamadan]

@nsf3ya Statistics does that to a lot of people. What chapter are you on?

[Hasan Ramadan @HasannRamadan]

A lot of statistics confusion is not the math.
It’s translating the output into a normal sentence.
What does this p-value mean for this question?
What should I check before I trust the model?
What would I tell a non-stats person?

That’s the part I help students and early analysts with.

[Hasan Ramadan @HasannRamadan]

@Mbadiwejesse This is a good map. The part students usually want help with is the layer under the dashboard: cleaning, choosing a test, and explaining the result.