KPIT Technologies Data Scientist Interview Questions and Answers

Q1. What is tokenization in NLP? and, to get raw tokens for a sentence with words seperated by space, why use tokenizers from nltk instead of str.split()?
Ans. 
Tokenization in NLP is the process of breaking down text into smaller units called tokens.
• Tokenization is a fundamental step in NLP for text preprocessing.

• Tokens can be words, phrases, or even individual characters.

• Tokenization helps in preparing text data for further analysis or modeling.

• NLTK tokenizers provide additional functionalities like handling contractions, punctuation, etc.

• str.split() may not handle complex t...read more
Answered by AI
View 1 more answer
Q2. While tokenzing, if you want to not break some word pairs(or triplets), for ex, to not tokenize the words 'first' and 'name' when they occur together and consider them as a single token, how would you do i...read more
Ans. 
Use NLTK's MWETokenizer to preserve specific word pairs or triplets during tokenization.
• MWETokenizer allows you to define multi-word expressions (MWEs) that should be treated as single tokens.

• Example: If you define MWE as ('first', 'name'), the tokenizer will keep 'first name' together.

• You can add multiple MWEs, such as ('New', 'York') and ('data', 'science').

• This is particularly useful in NLP tasks where context matte...read more
Answered by AI
Add your answer
Q3. For a data with 1000 samples and 700 dimensions, how would you find a line that best fits the data, to be able to extrapolate? this is not a supervised ML problem, there's no target. and how would you do i...read more
Ans. 
To find a line that best fits the data with 1000 samples and 700 dimensions, we can use linear regression.
• For unsupervised ML approach, we can use Principal Component Analysis (PCA) to reduce dimensions and then fit a line using linear regression.

• For supervised ML approach, we need to select a target column. We can choose any of the 700 dimensions as the target and treat it as a regression problem.

• Potential problems of...read more
Answered by AI
View 4 more answers

Q1. This round was entirely based on my projects on the resume!!!!
Add your answer
Q2. Explain project starting from the business problem to the deployment? [I was being interrupted for clarifications and questions]
Add your answer
Q3. [all of the follow up questions were baed on my answers] (some questions on businesss problem and rationale behing it, end users, how success will be measured and data collection that I can't disclose) a...read more
Add your answer

Q1. Gradient descent
Add your answer
Q2. Precision Recall
Add your answer

Q1. Where you see yourslef in next 5 year
Add your answer
Q2. Difficult situation which you face in ur last job
Add your answer

Q1. Questions related to the projects in the resume and some questions on machine learning concepts.
Add your answer

Q1. About the Projects I was working on and step by step detailing
Add your answer
Q2. Since it is NLP related projects mostly on Embeddings and word related questions
Add your answer

Its a take-home assignment related to NLP multi-class classification

Q1. Again they discussed my insights on the projects
Add your answer
Q2. Some case studies and how we can solve it
Add your answer

Q1. Salary Negotiation
Add your answer
Q2. Accepting offer
Add your answer

Q1. Basic DS question like how to handle missing features
Ans. 
Handling missing features involves techniques like imputation, deletion, and using algorithms that support missing values.
• Imputation: Replace missing values with mean, median, or mode. Example: If age is missing, use the average age of the dataset.

• Deletion: Remove rows or columns with missing values. Example: Drop a column if more than 50% of its values are missing.

• Using algorithms: Some algorithms like XGBoost can han...read more
Answered by AI
Add your answer

Q1. Case study based questions
Add your answer

Q1. How can Logistic regression be applied for multiclasstext classification
Ans. 
Logistic regression can be applied for multiclasstext classification by using one-vs-rest or softmax approach.
• One-vs-rest approach: Train a binary logistic regression model for each class, treating it as the positive class and the rest as the negative class.

• Softmax approach: Use the softmax function to transform the output of the logistic regression into probabilities for each class.

• Evaluate the model using appropriate...read more
Answered by AI
Add your answer

Q1. How does fbpropher forecasting model works and how is can be used to forecsst trafffic
Ans. 
fbprophet is a forecasting model developed by Facebook that uses time series data to make predictions.
• fbprophet is an open-source forecasting tool developed by Facebook's Core Data Science team.

• It is based on an additive model where non-linear trends are fit with yearly, weekly, and daily seasonality, plus holiday effects.

• fbprophet can be used to forecast traffic by providing historical data on traffic patterns and usi...read more
Answered by AI
Add your answer

Q1. Explain Decision Trees
Ans. 
Decision Trees are a popular machine learning algorithm used for classification and regression tasks.
• Decision Trees are a tree-like structure where each internal node represents a feature or attribute, each branch represents a decision rule, and each leaf node represents the outcome.

• They are easy to interpret and visualize, making them popular for exploratory data analysis.

• Decision Trees can handle both numerical and c...read more
Answered by AI
Add your answer
Q2. How will you finetune LLMs
Add your answer