N+A Data Scientist Interview Questions and Answers

Data Science MCQ questions

Building a baseline ML model with EDA etc.

Q1. How do you analyse outliers?
Ans. 
Outliers can be analyzed using statistical methods like Z-score, IQR, or visualization techniques like box plots.
• Calculate Z-score and identify data points with Z-score greater than a certain threshold as outliers.

• Use Interquartile Range (IQR) to detect outliers by identifying data points outside 1.5 * IQR range.

• Visualize data using box plots to identify any data points that fall outside the whiskers.

• Consider domain kn...read more
Answered by AI
Add your answer
Q2. Tell me about a time when what you found did not match expectation with what you thought analysing the dataset.
Add your answer

Q1. Why Unilever now?
Add your answer
Q2. Where do you see yourself in 5 years?
Add your answer

The first technical round will cover how computer vision works, including the advantages and disadvantages of regression and random forest. It will also include discussions on when to use precision and recall, methods to reduce false positives, and criteria for selecting different models. Additionally, disadvantages of PCA will be addressed, along with project-related questions. The second round will focus on standard aptitude tests, while the third round will involve a casual conversation with the Executive Vice President.

Normal aptitude questions

Machine learning and artificial intillegence

Q1. Metrics and related questions
Add your answer

Test was conducted on datacamp assessments. Overall, there were three tests.
1. Stats test
2. ML test
3. Python/coding test

Q1. Questions on ML techniques and practices, how to handle large data in python, lots of logical questions and handling overfitting, underfitting, etc in model building.
Add your answer

Q1. Machine learning algorithms.
Ans. 
Machine learning algorithms are tools used to analyze data, identify patterns, and make predictions without being explicitly programmed.
• Machine learning algorithms can be categorized into supervised, unsupervised, and reinforcement learning.

• Examples of machine learning algorithms include linear regression, decision trees, support vector machines, and neural networks.

• These algorithms require training data to learn patte...read more
Answered by AI
Add your answer
Q2. Credit risk life cycle
Add your answer
Q3. Pandas related questions
Add your answer

Q1. Steps of developing a credit risk model
Ans. 
Developing a credit risk model involves several steps to assess the likelihood of a borrower defaulting on a loan.
• 1. Define the problem and objectives of the credit risk model.

• 2. Gather relevant data such as credit history, income, debt-to-income ratio, etc.

• 3. Preprocess the data by handling missing values, encoding categorical variables, and scaling features.

• 4. Select a suitable machine learning algorithm such as logi...read more
Answered by AI
Add your answer
Q2. Pandas related questions
Add your answer
Q3. Bagging and Boosting
Add your answer

Q1. Explain AIC and BIC
Ans. 
AIC and BIC are statistical measures used for model selection in the context of regression analysis.
• AIC (Akaike Information Criterion) is used to compare the goodness of fit of different models. It penalizes the model for the number of parameters used.

• BIC (Bayesian Information Criterion) is similar to AIC but penalizes more heavily for the number of parameters, making it more suitable for model selection when the focus...read more
Answered by AI
Add your answer
Q2. Difference between xgboost and lightgbm
Ans. 
XGBoost is a popular gradient boosting library while LightGBM is a faster and more memory-efficient alternative.
• XGBoost is known for its accuracy and performance on structured/tabular data.

• LightGBM is faster and more memory-efficient, making it suitable for large datasets.

• LightGBM uses a histogram-based algorithm for splitting whereas XGBoost uses a level-wise tree growth strategy.
Answered by AI
Add your answer
Q3. Bagging and boosting
Add your answer

Q1. What is difference between lstm and rnn
Ans. 
LSTM is a type of RNN that addresses the vanishing gradient problem by using memory cells.
• RNN stands for Recurrent Neural Network, a type of neural network that processes sequential data.

• LSTM stands for Long Short-Term Memory, a type of RNN that includes memory cells to retain information over long sequences.

• LSTM is designed to overcome the vanishing gradient problem, which occurs when training RNNs on long sequences.

• L...read more
Answered by AI
Add your answer

Q1. What are different evaluation matrix ?
Ans. 
Evaluation matrices are used to assess the performance of models in data science.
• Confusion matrix: used to evaluate the performance of classification models.

• Precision, recall, and F1 score: measures for binary classification models.

• Mean squared error (MSE): evaluates the performance of regression models.

• R-squared: assesses the goodness of fit for regression models.

• Area under the ROC curve (AUC-ROC): evaluates the perfo...read more
Answered by AI
Add your answer