Statistics and Probability

Statistics and Probability Questions

Q1.Explain the Central Limit Theorem and why it matters.+

CLT states that the sampling distribution of the mean approaches a normal distribution as sample size increases, regardless of the population's distribution. It matters because it allows us to make inferences about population parameters from samples, use confidence intervals, and apply hypothesis tests. A sample size of 30+ is generally sufficient for CLT to apply.

Q2.What is a p-value and how do you interpret it?+

A p-value is the probability of observing results as extreme as the data, assuming the null hypothesis is true. If p < 0.05 (typical threshold), we reject the null hypothesis. A p-value is NOT the probability that the null hypothesis is true. Common misinterpretation: a low p-value does not prove the alternative hypothesis — it only suggests the data is unlikely under the null.

Q3.Explain the difference between Type I and Type II errors.+

Type I (false positive): rejecting a true null hypothesis — claiming an effect exists when it does not. Type II (false negative): failing to reject a false null hypothesis — missing a real effect. Reducing one increases the other. Alpha controls Type I (typically 0.05). Power (1 - beta) controls Type II (aim for 0.80+). In medical testing, Type II is more dangerous; in spam filtering, Type I annoys users.

Q4.What is Bayes theorem? Give a practical example.+

Bayes theorem: P(A|B) = P(B|A) * P(A) / P(B). It updates probability of an event based on new evidence. Example: a disease test has 99% sensitivity and 1% false positive rate. If 0.1% of the population has the disease, a positive test means only ~9% chance of actually having the disease — because false positives outnumber true positives in a rare disease. This is why understanding priors matters.

Q5.When would you use a t-test vs a chi-squared test vs ANOVA?+

T-test: compare means of two groups (paired or independent). ANOVA: compare means of 3+ groups (one-way or two-way). Chi-squared: test independence between categorical variables or goodness-of-fit for distributions. All assume certain conditions — t-test assumes normality (or large n), ANOVA assumes equal variances, chi-squared needs expected frequencies above 5.

Python and SQL

Python and SQL Questions

Q6.How would you use Pandas to find the top 5 customers by total purchase amount?+

df.groupby('customer_id')['amount'].sum().nlargest(5) or equivalently df.groupby('customer_id').agg(total=('amount', 'sum')).sort_values('total', ascending=False).head(5). This groups by customer, sums purchase amounts, sorts descending, and takes the top 5. In SQL: SELECT customer_id, SUM(amount) AS total FROM orders GROUP BY customer_id ORDER BY total DESC LIMIT 5.

Q7.Write a SQL query to find the second highest salary.+

SELECT MAX(salary) FROM employees WHERE salary < (SELECT MAX(salary) FROM employees). Or using window functions: SELECT salary FROM (SELECT salary, DENSE_RANK() OVER (ORDER BY salary DESC) as rank FROM employees) t WHERE rank = 2. DENSE_RANK handles ties correctly — if two employees share the highest salary, the next distinct salary is ranked 2nd.

Q8.How do SQL window functions work? Explain with ROW_NUMBER, RANK, and LAG.+

Window functions compute values across a set of rows related to the current row. ROW_NUMBER: unique sequential number per partition. RANK: same rank for ties, skips next (1,2,2,4). DENSE_RANK: same rank for ties, no skip (1,2,2,3). LAG(col, n): access value from n rows before. Example: SELECT employee_id, salary, LAG(salary) OVER (ORDER BY hire_date) as prev_salary — compares each employee's salary with the previously hired person.

Q9.How do you handle duplicate rows in a Pandas DataFrame?+

Identify: df.duplicated().sum() counts duplicates. df[df.duplicated()] shows duplicate rows. Remove: df.drop_duplicates() keeps first occurrence, df.drop_duplicates(keep='last') keeps last. Subset: df.drop_duplicates(subset=['email']) removes duplicates based on specific columns. Always investigate why duplicates exist before removing — they might indicate a data pipeline issue.

Q10.How would you merge two DataFrames in Pandas?+

pd.merge(df1, df2, on='key') for SQL-like joins. how parameter: 'inner' (matching rows), 'left' (all from df1), 'right' (all from df2), 'outer' (all from both). Use left_on/right_on when column names differ. pd.concat([df1, df2]) for stacking rows or columns. Always check for duplicate keys — one-to-many merges increase row count.

Want to master these concepts hands-on?

CodingClave's Data Science training in Lucknow covers all these topics with practical projects and mock interviews.

Learn Data Science at CodingClave

Machine Learning for Data Science

Machine Learning for Data Science Questions

Q11.How do you select features for a machine learning model?+

Filter methods: correlation matrix, mutual information, chi-squared test. Wrapper methods: recursive feature elimination (RFE), forward/backward selection. Embedded methods: L1 regularization (Lasso), tree-based feature importance. Also consider: remove zero-variance features, handle multicollinearity (VIF > 5), and use domain knowledge. Start simple — fewer features often perform better.

Q12.Explain how you would approach an A/B test.+

Define hypothesis and success metric. Calculate required sample size (power analysis). Randomly assign users to control (A) and treatment (B). Run test for sufficient duration (1-2 business cycles). Check for statistical significance using appropriate test (t-test for means, chi-squared for proportions). Check practical significance (effect size). Watch for novelty effects and segment analysis.

Q13.How do you communicate model results to non-technical stakeholders?+

Focus on business impact, not technical details. Use visualizations (charts, confusion matrices with plain labels). Express metrics in business terms: 'the model correctly identifies 92% of churning customers, allowing us to intervene before they leave.' Show dollar impact. Present confidence intervals. Acknowledge limitations. Use A/B test results as validation. Avoid jargon — say 'prediction accuracy' not 'F1-score'.

Problem Solving

Problem Solving Questions

Q16.A marketing team asks you to predict which customers will churn. Walk through your approach.+

Define churn (no purchase in 90 days?). Gather data: transaction history, demographics, engagement metrics. EDA: visualize churn rate, identify patterns. Feature engineering: recency, frequency, monetary value, engagement scores. Model: logistic regression (interpretable) or XGBoost (accuracy). Evaluate: precision-recall curve, lift chart. Deploy: weekly batch predictions, trigger retention campaigns for high-risk customers. Monitor churn rate reduction.

Q17.You notice your model's accuracy dropped from 95% to 88% in production. How do you diagnose?+

Check for data drift: compare feature distributions in production vs training data. Check for data quality issues: missing values, schema changes, upstream pipeline failures. Check for concept drift: the relationship between features and target may have changed (seasonality, market shifts). Compare model predictions on a held-out labeled set. If data drift: retrain on recent data. If quality: fix pipeline. If concept drift: update features or model architecture.

Frequently Asked Questions

Common Questions

What topics are asked in data science interviews?+

Statistics, probability, SQL, Python (Pandas, NumPy), ML algorithms, data visualization, and business case studies. Some companies include a take-home data analysis assignment.

Does CodingClave prepare students for data science interviews?+

Yes. The 6-month Data Science program at CodingClave in Lucknow includes mock interviews covering all data science interview formats.

How important is SQL for data science interviews?+

Very. Almost every data science interview includes a SQL round. Expect complex JOINs, window functions, CTEs, and data transformation queries.

What salary can I expect after clearing a data science interview?+

₹4-9 LPA for freshers. In Lucknow, ₹4-7 LPA. Companies like Fractal, MuSigma, and Tiger Analytics offer ₹6-12 LPA for analysts with strong data science skills.

Should I prepare case studies for data science interviews?+

Yes. Companies like consulting firms and analytics companies ask business case studies. Practice structuring problems: understand the business goal, define metrics, propose an analytical approach, and communicate expected impact.

Is Kaggle experience valued in data science interviews?+

Yes. Kaggle competitions demonstrate practical ML skills. A published kernel with good explanation shows both technical ability and communication skills.

How is a data science interview different from an ML engineer interview?+

Data science interviews emphasize statistics, EDA, business understanding, and communication. ML interviews emphasize model deployment, engineering, and scalability. Prepare for your target role specifically.

Do I need to memorize formulas for data science interviews?+

Understand concepts intuitively rather than memorizing formulas. Know what Bayes theorem does, when to use a t-test vs z-test, and how gradient descent works. Be able to explain in plain language.

Related Resources

Explore More Resources

PHP Laravel Interview Questions 2026

Top 15 PHP Laravel interview questions with answers for freshers and experienced...

Python Django Interview Questions 2026

Top 15 Python Django interview questions with answers for freshers. Prepare for ...

MERN Stack Interview Questions 2026

Top 15 MERN Stack interview questions with answers for freshers. MongoDB, Expres...

MEAN Stack Interview Questions 2026

Top 15 MEAN Stack interview questions with answers. Angular, Node.js, Express, M...

Get Started

Learn Data Science at CodingClave Lucknow

Join our practical-based Data Science training in Lucknow. 28-day, 45-day, and 6-month programs with projects, mentorship, and job assistance.

Apply Now WhatsApp Us View Data Science Course

Data Science Interview Questions and Answers 2026

Top 15 Data Science interview questions with answers. Statistics, ML, SQL, and Python questions for fresher data scientists in India.