Technical Interview Guide4 Practical Questions

Python Interview Questions for Data Analysts

Q: What is the operational difference between merge(), join(), and concat() in Pandas?

merge() performs SQL-style relational joins (INNER, LEFT, RIGHT, OUTER) between two DataFrames based on common key columns. join() combines DataFrames primarily on their index values. concat() stacks or appends DataFrames along an axis (axis=0 for stacking rows, axis=1 for side-by-side concatenation) without evaluating key matches.

Q: How does groupby() with .agg() work in Pandas, and why is it superior to manual loops?

groupby() splits a DataFrame into groups based on key columns, applies aggregate functions, and combines results into a new summary DataFrame. Using .agg() allows you to calculate multiple distinct metrics on different columns simultaneously in a single pass with C-optimized speed.

Python coding rounds evaluate whether you understand data structures, vectorization, and data wrangling with Pandas. Review these real interview coding questions tested across data analytics and business intelligence roles.

1Pandas Indexing

What is the fundamental difference between loc and iloc in Pandas?

What the Interviewer is Testing: Understanding label-based vs integer position-based data selection.

Model Answer:

loc is label-based: you select rows and columns by their index names or column labels, and slice endpoints are INCLUSIVE. iloc is integer position-based: you select rows and columns by their zero-indexed numeric positions (0, 1, 2...), and slice endpoints follow standard Python convention and are EXCLUSIVE.

Code Implementation:

# Label-based: inclusive
df.loc[0:5, ['customer_id', 'total_amount']]

# Position-based: exclusive (rows 0 to 4, columns 0 to 2)
df.iloc[0:5, 0:3]

Common Candidate Mistake: Using numeric slices in loc expecting exclusive stop indices, which unexpectedly includes an extra row in the dataset.

2Data Cleaning

How do you detect, evaluate, and handle missing values in a Pandas DataFrame?

What the Interviewer is Testing: Data preprocessing rigor and awareness of imputation bias.

Model Answer:

First evaluate null distribution with df.isna().sum(). Depending on business domain and percentage of missingness: 1) Drop rows or columns using df.dropna() if missing values are minimal (<2%) and random. 2) Impute numeric columns with median (if skewed) or mean using df["col"].fillna(df["col"].median()). 3) For time-series, use forward-fill (ffill). 4) For categorical features, impute with "Unknown" or the mode.

Code Implementation:

# Percentage of nulls per column
null_pct = df.isna().mean() * 100

# Impute skewed revenue with median
df['order_value'] = df['order_value'].fillna(df['order_value'].median())

Common Candidate Mistake: Blindly dropping all rows with null values using df.dropna(), which can eliminate 30%+ of valid business records without justification.

3Combining Data

What is the operational difference between merge(), join(), and concat() in Pandas?

What the Interviewer is Testing: Relational algebra and dataset integration in Python.

Model Answer:

merge() performs SQL-style relational joins (INNER, LEFT, RIGHT, OUTER) between two DataFrames based on common key columns. join() combines DataFrames primarily on their index values. concat() stacks or appends DataFrames along an axis (axis=0 for stacking rows, axis=1 for side-by-side concatenation) without evaluating key matches.

Code Implementation:

# Relational join on common key
merged_df = pd.merge(orders_df, customers_df, on='customer_id', how='left')

# Vertical stacking of monthly log files
annual_df = pd.concat([jan_df, feb_df, mar_df], axis=0, ignore_index=True)

Common Candidate Mistake: Using concat(axis=1) assuming it will match rows by customer_id, resulting in mismatched misaligned records if indices are not identical.

4Aggregation

How does groupby() with .agg() work in Pandas, and why is it superior to manual loops?

What the Interviewer is Testing: Vectorized split-apply-combine workflow.

Model Answer:

groupby() splits a DataFrame into groups based on key columns, applies aggregate functions, and combines results into a new summary DataFrame. Using .agg() allows you to calculate multiple distinct metrics on different columns simultaneously in a single pass with C-optimized speed.

Code Implementation:

summary = df.groupby('region').agg(
    total_orders=('order_id', 'count'),
    total_revenue=('amount', 'sum'),
    avg_order_value=('amount', 'mean')
).reset_index()

Common Candidate Mistake: Iterating through groups using for name, group in df.groupby(): ..., which is orders of magnitude slower than native vectorized .agg().

Live Technical Interview Preparation

Practice Live Technical Assessments with SSSAM Faculty

Reading questions is the first step. Writing queries under timed screen-share tests with live mentor reviews builds genuine interview confidence.

Explore Full Data Analyst Course →Book Free Demo →

SQL Skill GuideMaster database queries, multi-table joins, and window functions →Portfolio ProjectsBuild practical case studies to present during interview rounds →2026 RoadmapThe full learning sequence to prepare for analyst hiring →