Python Data Analysis Examples

In the last lesson, Data Cleaning Basics, you learned to fix messy data. So now you have all the pieces: load a file, clean it, select and filter it. The question is, how do they fit together on a real job? That is what this lesson is. We take a small dataset of orders and answer real business questions about it, start to finish. By the end you will see the full rhythm of a data analysis, and meet the one tool that ties it all together: groupby.

🤔 What does “analyzing data” actually mean?

Analysis sounds fancy, but it just means asking questions of your data and getting answers backed by numbers. Not guessing. Counting.

A real analysis usually answers questions like:

• How much did we sell in total?
• Which product sells the most?
• What is the average order value per city?
• Who are our top customers?

Every one of these is just a count, a sum, or an average over a group of rows. Pandas does all of them in a line or two. Let’s see it on a real table.

🧱 Our dataset

Here is a small table of orders. In real life this would come from a CSV with pd.read_csv, but we will build it directly so you can run the whole lesson without a file.

import pandas as pd

data = {
    "order_id": [1, 2, 3, 4, 5, 6, 7, 8],
    "customer": ["Alex", "Riya", "Alex", "Maria", "Riya", "Sam", "Alex", "Maria"],
    "product": ["Pen", "Notebook", "Pen", "Bag", "Pen", "Notebook", "Bag", "Pen"],
    "city": ["London", "Mumbai", "London", "Toronto", "Mumbai", "London", "London", "Toronto"],
    "amount": [50, 120, 50, 800, 50, 120, 800, 50]
}

df = pd.DataFrame(data)
print(df.head())

   order_id customer   product     city  amount
0         1     Alex       Pen   London      50
1         2     Riya  Notebook   Mumbai     120
2         3     Alex       Pen   London      50
3         4    Maria       Bag  Toronto     800
4         5     Riya       Pen   Mumbai      50

So each row is one order: who bought it, what product, in which city, and the amount. Now let’s ask questions.

💰 Simple totals and counts

The easiest questions are about the whole table. You sum a column, or count its rows, in one call.

This answers “how much did we sell” and “how many orders”:

print("Total sales:", df["amount"].sum())
print("Number of orders:", len(df))
print("Average order:", df["amount"].mean())
print("Biggest order:", df["amount"].max())

Total sales: 1990
Number of orders: 8
Average order: 248.75
Biggest order: 800

Each line is one question:

• df["amount"].sum() adds up every order amount, giving total sales.
• len(df) counts the rows, which is the number of orders.
• .mean() and .max() give the average and the largest order.

These whole-table numbers are a good first look. But the real value comes from breaking the data into groups.

📊 Grouping with groupby

Here is the most important tool in data analysis. groupby splits the table into groups that share a value, then runs a calculation on each group. It answers “per something” questions, like sales per city or orders per customer.

This finds the total sales for each city:

sales_by_city = df.groupby("city")["amount"].sum()
print(sales_by_city)

city
London     1700
Mumbai      170
Toronto     1600
Name: amount, dtype: int64

Read the line left to right, because that is exactly how it works:

• df.groupby("city") splits the rows into groups, one per city.
• ["amount"] picks the amount column inside each group.
• .sum() adds that column up for each group.

So you get one total per city, all at once. London sold 1700, Mumbai 170, Toronto 1600. You can swap .sum() for .mean() to get the average per group, or .count() to count rows per group. The pattern is always the same: group, pick a column, calculate.

This shows the average order amount per customer instead:

avg_by_customer = df.groupby("customer")["amount"].mean()
print(avg_by_customer)

customer
Alex     300.000000
Maria    425.000000
Riya      85.000000
Sam      120.000000

Same pattern, different question. We grouped by customer this time and asked for the average. That is the power of groupby: change the column or the calculation and you answer a brand new question with almost the same line.

🔢 Counting categories with value_counts

A very common question is “how many of each?” For that, value_counts counts how often each value appears in a column.

This counts how many orders each product had:

print(df["product"].value_counts())

product
Pen         4
Bag         2
Notebook    2
Name: count, dtype: int64

So Pen was ordered 4 times, Bag and Notebook twice each. value_counts also sorts by the highest count first, so the most common item is right at the top. It is the fastest way to see what is popular.

⬆️ Sorting the results

Once you have numbers, you usually want them in order, biggest first. sort_values reorders the rows by a column.

This sorts the city totals from highest to lowest:

sales_by_city = df.groupby("city")["amount"].sum()
ranked = sales_by_city.sort_values(ascending=False)
print(ranked)

city
London     1700
Toronto    1600
Mumbai      170
Name: amount, dtype: int64

The key is ascending=False, which means largest first. Without it, sorting goes smallest first. Now you can instantly see London is the top city and Mumbai the lowest. Sorting turns a pile of numbers into a clear ranking.

🖥️ Putting it all together

Here is a complete mini-analysis that answers several questions in one script. This is the real flow you would write on the job.

import pandas as pd

data = {
    "customer": ["Alex", "Riya", "Alex", "Maria", "Riya", "Sam", "Alex", "Maria"],
    "product": ["Pen", "Notebook", "Pen", "Bag", "Pen", "Notebook", "Bag", "Pen"],
    "city": ["London", "Mumbai", "London", "Toronto", "Mumbai", "London", "London", "Toronto"],
    "amount": [50, 120, 50, 800, 50, 120, 800, 50]
}
df = pd.DataFrame(data)

print("Total sales:", df["amount"].sum())
print("\nTop product:")
print(df["product"].value_counts().head(1))
print("\nSales per city (high to low):")
print(df.groupby("city")["amount"].sum().sort_values(ascending=False))
print("\nTop customer by spend:")
print(df.groupby("customer")["amount"].sum().sort_values(ascending=False).head(1))

Total sales: 1990

Top product:
product
Pen    4
Name: count, dtype: int64

Sales per city (high to low):
city
London     1700
Toronto    1600
Mumbai      170
Name: amount, dtype: int64

Top customer by spend:
customer
Alex    900
Name: amount, dtype: int64

Look at what you just did with a handful of lines. You found total sales, the most-ordered product, a city ranking, and the top customer. No spreadsheets, no manual counting. That is data analysis: ask a question, write a short Pandas line, read the answer.

🛠️ Practice Challenge

Try these on the same df. Each one is one or two lines of Pandas. Give it a real attempt before opening the answer.

Question 1: How many orders did each customer place? (Hint: count rows per customer.)

print(df["customer"].value_counts())

customer
Alex     3
Maria    2
Riya     2
Sam      1
Name: count, dtype: int64

Question 2: What is the total amount sold for each product, ranked from highest to lowest?

result = df.groupby("product")["amount"].sum().sort_values(ascending=False)
print(result)

product
Bag         1600
Notebook     240
Pen          200
Name: amount, dtype: int64

⚠️ Common Mistakes

A few traps in analysis:

• Forgetting to pick a column after groupby. df.groupby("city").sum() tries to sum every column, which can include ones that make no sense. Pick the column you mean: df.groupby("city")["amount"].sum().
• Reading an unsorted result as a ranking. groupby returns groups in label order, not by size. Add sort_values(ascending=False) when you want a ranking.
• Confusing value_counts and groupby. Use value_counts to count how often each value appears. Use groupby when you want to sum or average another column per group.

✅ Best Practices

Habits for clean analysis:

• Always clean the data first (blanks, duplicates, types) before you trust any total or average.
• Build the analysis one question at a time. Print each result before moving to the next.
• Name your results clearly, like sales_by_city, so the code reads like the questions you are asking.
• Remember the group, pick, calculate, sort pattern. Most “per something” questions follow it exactly.

🧩 What You’ve Learned

✅ Analysis means asking questions of your data and answering them with counts, sums, and averages.

✅ Whole-table numbers come from df["col"].sum(), .mean(), .max(), and len(df).

✅ groupby splits the table into groups and calculates per group, the heart of “per something” questions.

✅ value_counts counts how often each value appears, and sort_values(ascending=False) ranks results biggest first.

✅ A real analysis is just these small steps in a row: load, clean, group, calculate, sort, read the answer.

🚀 What’s Next?

You can now take raw data from a file all the way to real answers. That is the full data analysis skill. From here, the next big step is teaching the computer to find patterns on its own. That is machine learning, and it starts with understanding how Python powers the whole world of AI.

Python in AI