Extract data

We rarely use all the data at once. Instead, we extract a relevant subset: perhaps transactions over $10 million, or only those involving a specific country. This allows us to focus on the question we’re trying to answer. Extracting data is like zooming in on one part of a graph — it gives clarity and insight.

In this section, you'll learn how to create new tables based on a condition, such as value thresholds, country names, or transaction types.

Extract rows based on one condition

Excel

In Excel, you use filters or conditional formatting to show only rows that meet a certain condition.

t0 Prompt

Give me all rows where the amount is over 10 million

Extract transactions from Sweden

Show only goods transactions

Code

The python code looks as follows:

transactions["Amount"] = pd.to_numeric(transactions["Amount"])
high_value = transactions[transactions["Amount"] > 10_000_000]
high_value

Extract rows based on multiple conditions

Excel

In Excel, you might apply filters to multiple columns or use formulas like =AND() or =OR().

t0 Prompt

Show me rows where amount is over 10 million and buyer country is Brazil

Extract transactions that are services and from Italy

Filter column X by condition Z and Y

Code

The python code looks as follows:

transactions["Amount"] = pd.to_numeric(transactions["Amount"])
subset = transactions[
    (transactions["Amount"] > 10_000_000) &
    (transactions["Buyer Country"] == "Brazil")
]
subset

Extract specific columns only

Excel

In Excel, you might copy specific columns into a new worksheet to focus your analysis.

t0 Prompt

Give me only the date, amount, and seller

I want to work with a smaller table

Extract columns X and Y

Code

The python code looks as follows:

small_table = transactions[["Date", "Amount", "Seller"]]
small_table