Mu Sigma Interview Questions with Answers
Last updated 2026-11-19
Mu Sigma is a decision-science and analytics firm, and the preparation advice you will find for it centres on structured thinking: aptitude, logic puzzles, guesstimates and small business cases, with only light coding for most entry-level tracks. The point of such questions is to watch how you break an unfamiliar problem into parts and communicate your reasoning.
These twelve practice questions come with worked answers and are written in the style of those topics, not as a record of any interview. For more numerical and SQL practice, continue with Mu Sigma coding round questions and the data analyst interview guide.
| Rounds (commonly reported) | Aptitude test, technical or analytical interview, HR |
|---|---|
| Topics | Guesstimates, puzzles, probability, business cases, basic SQL |
| Roles | Decision science and analytics tracks |
Rounds as commonly reported
We hold no sourced description of the company's current selection stages, so treat this as general guidance drawn from how analytics consultancies assess freshers. Candidates commonly describe an aptitude screen (quantitative, logical, verbal), an interview that mixes puzzles, a guesstimate or case, and some technical questions on statistics or SQL, then an HR conversation.
- Aptitude: quantitative ability, data interpretation, logical reasoning, verbal.
- Analytical interview: a guesstimate, a puzzle, a case, basic statistics, sometimes SQL.
- HR: motivation, adaptability, location and expectations.
How to think aloud
There is rarely one right number, so interviewers watch your structure. Say what you will estimate, split it into factors you can reason about, state each assumption, compute with round numbers and finish by sanity-checking against a second approach. Silence for a minute while you think is fine if you say you are structuring.
Hiring status
We hold no sourced hiring announcements, drive dates, openings or process description for Mu Sigma, so none are stated here. It has no page in our hiring section. Check the company's own careers site for current roles and eligibility.
Mu Sigma Interview Questions with Answers: questions with worked answers
1 Tell me about yourself and why decision science.
easyGive a short background, then connect it to the work: you enjoy turning ambiguous questions into numbers, and you can point to one example, a class project or a hackathon where you cleaned data and made a recommendation. Say what you want to learn: structured problem solving, tools such as SQL, Python or R, and communicating insight to a business audience. Avoid claiming deep model expertise you cannot demonstrate.
2 Guesstimate: how many petrol pumps are there in Bengaluru?
mediumPick an approach and say so. Demand side: assume about 13 million people and about 0.4 vehicles per person gives roughly 5 million vehicles, treating two-wheelers and cars together. Suppose each vehicle refuels about once every 5 days: 1 million refuelling visits per day. If a pump serves about 1,000 vehicles per day (roughly 40 per hour over a day, with peaks), the city needs about 1,000 pumps. Cross-check supply side: if a pump covers roughly a 2 km by 2 km catchment and the built-up area is about 700 square kilometres, you get about 175 geographic cells, which suggests multiple pumps per cell, so 700 to 1,000 is plausible. State that the assumptions are yours and would need data.
3 Puzzle: there are 8 balls, one heavier. Using a balance, find it in two weighings.
mediumSplit into groups of 3, 3 and 2. Weigh 3 against 3. If they balance, the heavy ball is in the group of 2: weigh those two, and the heavier side is the ball. If one side is heavier, it holds the heavy ball: from those 3, weigh any two against each other; the heavier one is the ball, and if they balance it is the third. Either path takes exactly two weighings. The idea is that each weighing has three outcomes, so two weighings can distinguish up to 9 cases.
4 Puzzle: three switches outside a room control three bulbs inside. You may enter once. How do you match them?
mediumTurn switch 1 on and wait ten minutes, then turn it off. Turn switch 2 on and enter. The bulb that is lit belongs to switch 2. Of the two unlit bulbs, the one that is warm belongs to switch 1 (it was on and heated up), and the cold one belongs to switch 3. It works because heat gives you a third observable state in addition to on and off.
5 Two fair dice are thrown. What is the probability the sum is 7, and the probability of at least one six?
mediumThere are 36 equally likely outcomes. A sum of 7 occurs for (1,6), (2,5), (3,4), (4,3), (5,2), (6,1): 6 outcomes, so 6/36 = 1/6. At least one six: use the complement. No six has 5 x 5 = 25 outcomes, so at least one six is 1 - 25/36 = 11/36. Always check by listing when the sample space is small.
pythonfrom itertools import product rolls = list(product(range(1, 7), repeat=2)) print(sum(a + b == 7 for a, b in rolls) / 36) # 0.1666... print(sum(6 in r for r in rolls) / 36) # 0.3055...6 A disease affects 1 percent of people. A test detects it in 90 percent of those who have it and falsely flags 9 percent of those who do not. You test positive. What is the chance you have it?
mediumUse Bayes' theorem. Per 10,000 people: 100 have the disease, of whom 90 test positive; 9,900 do not, of whom 9 percent, or 891, test positive. Total positives are 981, and 90 of them are true, so the probability is 90/981, about 9.2 percent. Most positives are false because the disease is rare. This base-rate effect is the point of the question.
7 Case: a telecom operator's customer churn has risen. How would you approach it?
hardDefine churn (voluntary versus non-payment) and the period, then see where it rose: by plan, region, tenure and acquisition channel. Hypotheses: a competitor offer, network quality problems, a price change, poor service experience. Data to check: usage trends before leaving, complaint counts, billing shocks and cohort retention. Propose actions matched to cause: targeted retention offers to high-value at-risk users, fixing coverage in affected areas, a proactive call after a bad experience. Define success as a lower churn rate at acceptable cost per retained customer, tested with a control group.
8 When do you use mean, median and mode? Give an example where they differ.
mediumThe mean uses all values and is sensitive to outliers. The median is the middle value and resists outliers, so it is the better summary for skewed data such as salaries or house prices. The mode is the most frequent value and suits categorical data such as the most common shoe size. For incomes 20, 22, 25, 28 and 500 thousand, the mean is 119 thousand, the median 25 thousand, and the mean badly misdescribes a typical person.
pythonimport statistics as st x = [20, 22, 25, 28, 500] print(st.mean(x), st.median(x)) # 119 259 Explain correlation versus causation with an example.
mediumCorrelation measures how two variables move together; causation means one produces a change in the other. Ice-cream sales and drowning incidents are correlated because both rise in summer, a confounder. To argue causation you need a randomised experiment, or careful observational designs that control confounders, plus a plausible mechanism and a consistent time order. Report correlations as associations unless you have that evidence.
10 A price rises by 20 percent and then falls by 20 percent. What is the net change?
easyTake the price as 100. After the rise it is 120; a 20 percent fall from 120 removes 24, leaving 96. The net change is a 4 percent decrease. Percentage changes are not additive because the base changes: a gain of x percent followed by a loss of x percent always leaves a net loss of x squared over 100 percent, here 400 / 100 = 4.
11 Write a SQL query to find duplicate emails in a users table.
mediumGroup by the email and keep groups with a count above one. Normalise case and spaces if "A@x.com" and "a@x.com " should match. To see the full duplicate rows, join back or use a window COUNT. Mistake: using DISTINCT, which hides duplicates rather than finding them.
sqlSELECT LOWER(TRIM(email)) AS email, COUNT(*) AS n FROM users GROUP BY LOWER(TRIM(email)) HAVING COUNT(*) > 1;12 Describe a time you worked on a problem that had no clear instructions.
mediumChoose a real instance. Say how you clarified the goal by asking one or two questions, broke the work into steps you could finish, made your assumptions explicit and checked them with someone, and what you delivered. End with what you learnt about handling ambiguity. Decision-science employers like candidates who make progress without waiting for perfect instructions, and who flag assumptions.
Frequently asked questions
What does a Mu Sigma interview focus on?
Preparation sources describe aptitude, guesstimates, puzzles, small business cases and basic statistics or SQL. We hold no sourced company description, so confirm the stages with your recruiter.
How do I prepare for guesstimates?
Practise breaking a quantity into multiplied factors, using round numbers, stating assumptions and cross-checking with a second approach. Do one a day for two weeks.
Is coding required?
For many analytics tracks, light SQL and basic Python or R are enough, but check the role description. Prepare SQL joins and aggregation.
Where do I find current openings?
On the company's own careers site. We hold no sourced hiring dates for Mu Sigma.
Are these real Mu Sigma questions?
No, they are practice questions in the style of such interviews, not a record of any interview.
Related guides
More in Company Interview Questions
- Myntra Coding Round Questions with Solutions
- Myntra Interview Questions and Answers
- Nagarro coding round questions
- Nagarro interview questions and answers
- Netflix Coding Round Questions with Solutions in Python and Java
- Netflix Interview Questions and Answers for Engineers
- NTT DATA coding round questions
- NTT DATA interview questions and answers
- Nutanix Coding Round Questions With Solutions
- Nutanix Interview Questions and Answers
- NVIDIA Coding Round Questions with Solutions
- NVIDIA Interview Questions and Answers for India Software Roles