Practice pandas interview questions with a duplicate-key merge example, missing-data decisions, groupby reasoning and checks that protect report accuracy. Start with the question, explain the mechanism, and then state an assumption or tradeoff. The scenarios below are original practice examples, not questions supplied by an employer.
What is the difference between merge and concat?
A merge matches rows using keys; concatenation combines objects along an axis. State whether you are adding more observations, adding aligned columns or matching records from different tables. Before combining data, describe the expected row grain. Two tables can share a column name while representing very different entities. A useful answer includes the expected row count and a validation step, not just the method name.
Why can a merge unexpectedly multiply rows?
Repeated keys on both sides can produce multiple matches for each key. If two order rows share a customer ID and three customer-history rows share it, joining on that ID creates six matched rows. This may be correct for a history analysis but wrong for a report requiring one current customer record. Check key uniqueness and choose the right relationship instead of removing duplicates from the result without investigating their meaning.
How should you handle missing values?
Ask why the value is missing and what the calculation means. A missing sales amount may represent an incomplete import, while zero may represent a valid transaction with no charge. Treating both as zero can conceal a data-quality failure. Report missingness, distinguish optional from required fields and decide whether to exclude, impute or reject rows. Explain the consequence of the choice for totals and comparisons.
What does groupby change about the data?
Grouping moves from individual observations to summaries defined by the grouping keys. Explain the output grain explicitly: one row per customer, region or day. Choose aggregation functions that match the question; an average of daily averages need not equal an overall average. Retain counts and denominators when the result will be combined again. Check whether missing grouping keys should be included in the business report.
How do you make a transformation trustworthy?
Verify input types, required columns, identifier uniqueness and plausible ranges before transformation. Afterward, check row counts, unmatched records and control totals. Keep a small example whose expected output can be calculated by hand. Avoid optimizing a pipeline before confirming its meaning. If the data is large, discuss selecting needed columns and appropriate storage, but separate performance decisions from correctness checks.
Worked example
Original example: orders contain customer A twice, with amounts 10 and 20. A customer table contains A twice because two address versions were retained. A merge on customer ID produces four rows and a total of 60, even though the order total is 30.
The correction depends on the requirement. For a current-customer report, resolve which customer record is current and validate a many-to-one join. For an as-of report, include the relevant time relationship. Do not merely divide the result by two: a different customer may have three history records, and that shortcut would produce another wrong total.
Practice plan
Build the two tiny tables from the worked example and predict the merged row count before running it. Then introduce an unmatched customer and a missing amount. Explain how each should appear in an audit report. Rehearse your answer as a business explanation before mentioning implementation syntax.
Use Cluegent during preparation to review your own answer: ask for one incorrect assumption and one follow-up question, then respond again without suggestions. Check current plans before choosing a subscription. Follow the employer's rules during the actual interview.
Sources checked
These official references support the guide. Product details and technical documentation can change; check the linked source for current information.
Where Cluegent helps
Cluegent supports permitted live workflows with transcript context, typed prompts, screenshot-aware answers, resume context, custom response behavior, quick action buttons, and a private desktop overlay. It is most useful when you already understand the subject and need help staying structured under pressure.
Frequently asked questions
Should I use drop_duplicates after every merge?
No. First establish the intended row grain and why duplicates exist. Removing rows can hide a broken relationship or discard valid observations.
Are code-only answers enough?
For analyst work, explain the business definition, expected output and validation as well as the pandas operation.