Top 10 Python Questions Asked in Data Analyst Interviews (2026)

We aggregated 450+ interview transcripts submitted by MockMate AI users interviewing for data-analyst roles in the last 6 months. Below are the 10 Python questions that repeated the most — with a quick answer, and the common trap.

1. 1What is the difference between list and tuple? Lists are mutable, tuples aren’t. Trap: don’t stop there — add “tuples are hashable, so they can be dict keys”.
2. 2What is the GIL and why does it matter for a data analyst? Global Interpreter Lock — CPython runs one thread at a time. For data work it means multi-processing scales your pandas jobs, threading usually doesn’t.
3. 3Explain merge vs join vs concat in pandas. merge is SQL-style joining on keys, join is index-based and syntactic sugar over merge, concat stitches by axis.
4. 4How do you handle NaN in a large dataframe? Prefer fillna() with a domain-specific default, or dropna() only when the column is <5% missing. Trap: never blindly fill with zero for a currency column.
5. 5Difference between apply, map, and applymap? Series: map. DataFrame column-wise: apply. Every cell: applymap. Vectorised numpy is usually 10x faster than all three — mention this.
6. 6What does @staticmethod vs @classmethod do? staticmethod ignores self and cls, classmethod receives cls (the class itself). Use classmethod for alternate constructors.
7. 7Deep copy vs shallow copy — which one does df.copy() do? By default deep=True, so it’s a deep copy. Setting deep=False only copies the metadata.
8. 8How do you optimise a pandas group-by on 10M rows? Use categorical dtypes for group keys, sort=False, and prefer .agg() with numpy functions over lambdas.
9. 9What are Python generators — give a data example. Lazy iterators. Reading a 5 GB CSV in chunks with pd.read_csv(..., chunksize=100_000) yields a generator — you never load the whole file.
10. 10Write a one-liner to find the second-highest salary in a dataframe. df.nlargest(2, ‘salary’).iloc[-1]. Trap: don’t say sort_values().iloc[1] unless you handle duplicates.