Top Interview Questions
✨ Introduction
Are you preparing for Data Science or AI interviews in 2025? Whether you are a fresher stepping into the world of machine learning or an experienced developer polishing your skills, this guide is for you. In this post, we cover the Top 25 Python, NumPy & Pandas Interview Questions and Answers — explained with clear definitions, detailed explanations, and practical examples. Python is the backbone of AI, ML, and Data Science, while NumPy and Pandas are the most widely used libraries for data analysis and manipulation. These top interview questions will not only help you build strong concepts but also boost your confidence to crack technical interviews with ease.1. What is Python and why is it popular in AI/ML?
Answer: Python is a high-level, interpreted, and dynamically typed programming language known for its simplicity, readability, and extensive library support. Why Python is Popular in AI/ML Python has become the primary language for Artificial Intelligence (AI) and Machine Learning (ML) because:- Ease of Learning & Readability
- Python’s syntax is simple and close to natural language, making it easy for developers and researchers to write and understand code.
- This reduces development time and allows quick prototyping.
- Extensive Libraries & Frameworks
- Pre-built libraries make complex tasks easier:
- NumPy & Pandas – Data manipulation & numerical computation.
- Matplotlib & Seaborn – Data visualization.
- Scikit-learn – Machine learning algorithms.
- TensorFlow, PyTorch, Keras – Deep learning and neural networks.
- Pre-built libraries make complex tasks easier:
- Large Community Support
- Python has a massive global community that actively contributes tutorials, documentation, and open-source projects.
- Easier to find solutions to problems.
- Platform Independence
- Runs on Windows, macOS, Linux, and even embedded systems.
- Integration Capabilities
- Can easily integrate with C/C++, Java, R, and cloud services.
- Strong Support for Data Handling
- Works well with structured and unstructured data — critical in AI/ML projects.
2. What is NumPy?
Answer: NumPy (Numerical Python) is a Python library used for scientific computing, numerical analysis, and data manipulation. It provides powerful tools to work with multi-dimensional arrays (ndarrays), along with functions for mathematical, logical, statistical, and linear algebra operations. NumPy is the foundation for many AI/ML, Data Science, and scientific libraries like Pandas, SciPy, and scikit-learn. Key Features of NumPy- N-dimensional array object (ndarray) – Fast, memory-efficient, and easy to manipulate.
- Mathematical functions – Vectorized operations for speed.
- Broadcasting – Allows arithmetic between arrays of different shapes.
- Linear algebra & statistics – Supports matrix multiplication, eigenvalues, standard deviation, etc.
- Integration – Works well with Python, C/C++, and Fortran code.
- AI/ML involves large datasets and matrix operations (e.g., in neural networks).
- NumPy allows fast computation by using C-based implementations under the hood.
- Libraries like TensorFlow and PyTorch are inspired by NumPy’s array structure.
3. What is Pandas?
Answer: Pandas (short for Panel Data) is a Python library used for data manipulation, cleaning, and analysis. It provides two main data structures:- Series – 1D labeled array (like a column in Excel)
- DataFrame – 2D labeled data structure (like a spreadsheet or SQL table)
- DataFrames & Series – Store and manipulate structured data easily.
- Data Cleaning – Handle missing data (NaN), duplicate removal, and type conversions.
- Data Selection & Filtering – Use labels (.loc) or index positions (.iloc).
- Data Aggregation – Summarize with functions like groupby(), mean(), sum().
- File I/O – Read/write data from CSV, Excel, JSON, SQL, etc.
- Time Series Handling – Work with dates, timestamps, and resampling.
- AI/ML projects start with data preprocessing. Pandas makes data cleaning, transformation, and analysis very fast.
- Works seamlessly with NumPy arrays and integrates well with scikit-learn, TensorFlow, and PyTorch.
- Used for feature engineering, EDA (Exploratory Data Analysis), and data wrangling.
4. What is a DataFrame in Pandas?
Answer: In Pandas, a DataFrame is a two-dimensional, labeled data structure, similar to a table in a spreadsheet or a SQL database.- It has rows and columns.
- Each column can store data of different types (e.g., integers, floats, strings).
- Both rows and columns have labels (index for rows, column names for columns).
5. Explain Broadcasting in NumPy.
Answer: Broadcasting in NumPy is a powerful mechanism that allows arithmetic operations between arrays of different shapes.- Instead of manually reshaping or looping over arrays, NumPy automatically expands the smaller array to match the shape of the larger array.
- This makes computations faster and more memory-efficient.
- If the dimensions are equal, no problem.
- If one of the dimensions is 1, NumPy stretches it to match the other dimension.
- If dimensions are unequal and neither is 1, broadcasting fails.
6. How do you handle missing data in Pandas?
Answer: In Pandas, missing data (or null values) are typically represented as NaN (Not a Number) or None. Handling missing data is crucial in AI/ML because algorithms often cannot process null values. Pandas provides built-in methods to detect, remove, or fill missing data efficiently. Common Methods to Handle Missing Data- Detect Missing Data
7. What is a Series in Pandas?
Answer: In Pandas, a Series is a one-dimensional labeled array capable of holding any data type (integers, floats, strings, Python objects, etc.).- It is similar to a single column in a spreadsheet or a list with an index.
- Each element in a Series has a label (index) that allows for fast data access.
- 1D Structure – Contains a single column of data.
- Labeled Index – Each element has an associated index label.
- Heterogeneous Data – Can hold mixed data types (though typically homogeneous for numerical operations).
- Vectorized Operations – Supports element-wise arithmetic and operations efficiently.
8. Difference between loc[] and iloc[]?
Answer: loc[] (Label-based indexing)- Used to access rows and columns by labels (names/index values).
- Think: “loc means by name/label.”
- Used to access rows and columns by integer position (like array index).
- Think: “iloc means by index number.”
| Feature | loc[] (Label-based) | iloc[] (Integer-based) |
| Access type | By label (row/col names) | By integer index (0,1,2) |
| Row Example | df.loc[‘b’] | df.iloc[1] |
| Column Example | df.loc[:, ‘Age’] | df.iloc[:, 1] |
| Range Slicing | Inclusive (df.loc[‘a’:’b’] includes both) | Exclusive (df.iloc[0:2] stops before 2) |
| Use case | Human-readable queries | Programming / position-based |
9. What is vectorization?
Answer: Vectorization is the process of performing operations on entire arrays or sequences of data at once, rather than using explicit loops (like for or while loops).- It leverages low-level optimized C/Fortran code behind libraries like NumPy, making computations faster and more efficient.
- Commonly used in NumPy, Pandas, and AI/ML workflows for mathematical operations, data manipulation, and model computations.
- Performance Improvement – Eliminates Python-level loops, which are slow for large datasets.
- Readable Code – Allows concise, expressive code instead of nested loops.
- Supports Large Datasets – Efficiently handles arrays, matrices, and tensors for ML/AI tasks.
- Foundation for Libraries – Libraries like TensorFlow and PyTorch rely heavily on vectorized operations.
10. Explain data normalization in Python.
Answer: Data normalization is the process of scaling numerical features in a dataset to a common range, typically between 0 and 1 or -1 and 1.- It ensures that all features contribute equally to model training.
- Particularly important in AI/ML, where algorithms like gradient descent, KNN, SVM, and neural networks are sensitive to feature scale.
- Prevents bias – Features with larger values do not dominate.
- Speeds up convergence – Algorithms like gradient descent converge faster on normalized data.
- Improves accuracy – Helps distance-based algorithms (like KNN) work properly.
- Consistency – Ensures stable training for deep learning models.
11. What is one-hot encoding?
Answer: One-hot encoding is a technique used to convert categorical variables into a binary (0 or 1) representation so that they can be used in machine learning algorithms.- Each category in a feature is represented by a separate column.
- The column corresponding to the category has a 1, and all other columns have 0.
12. How to merge DataFrames?
Answer: In Pandas, merging DataFrames means combining two or more DataFrames based on a common column or index, similar to SQL JOIN operations. Pandas provides the function pd.merge() to merge DataFrames. Syntax: pd.merge(left, right, how=’inner’, on=None, left_on=None, right_on=None)- left, right → DataFrames to merge
- how → type of join (‘inner’, ‘outer’, ‘left’, ‘right’)
- on → column(s) to join on
- left_on, right_on → if column names differ in DataFrames
13. What is the use of apply() in Pandas?
Answer: The apply() function in Pandas is used to apply a function along the axis of a DataFrame (rows or columns) or to elements of a Series.- It is very powerful because it allows you to use custom functions (user-defined or lambda functions) on your data.
- Think of it as a way to loop through rows or columns in a smarter and faster way.
- func → function to apply
- axis=0 → apply function to each column (default)
- axis=1 → apply function to each row
14. What are lambda functions?
Answer: A lambda function in Python is a small, anonymous function defined using the keyword lambda instead of def.- Anonymous → it doesn’t need a name.
- Usually written in one line.
- Often used for short, simple operations.
- lambda → keyword.
- arguments → input values.
- expression → operation to perform (must return a value).
15. How to filter data in Pandas?
Answer: Filtering data in Pandas means selecting rows (or sometimes columns) from a DataFrame based on specific conditions (like greater than, equal to, contains text, etc.). Creating a Sample DataFrame import pandas as pd data = { ‘Name’: [‘Alice’, ‘Bob’, ‘Charlie’, ‘David’, ‘Eva’], ‘Age’: [25, 30, 35, 40, 22], ‘City’: [‘New York’, ‘Paris’, ‘London’, ‘Berlin’, ‘Paris’] } df = pd.DataFrame(data) print(df) ✅ Output: Name Age City 0 Alice 25 New York 1 Bob 30 Paris 2 Charlie 35 London 3 David 40 Berlin 4 Eva 22 Paris Filtering with a Condition Example: Age greater than 30 filtered_df = df[df[‘Age’] > 30] print(filtered_df) ✅ Output: Name Age City 2 Charlie 35 London 3 David 40 Berlin16. What are NumPy axis?
Answer: In NumPy, an axis refers to a particular dimension of an array along which operations are performed.- A 1D array has 1 axis → like a line.
- A 2D array has 2 axes → rows and columns.
- A 3D array has 3 axes → like a cube (depth, rows, columns).
-
- Only 1 axis → Axis 0
- Length = 4 → elements are along Axis 0.
- Axis-0 → rows (vertical ↓)
- Axis-1 → columns (horizontal →)
| Array Type | Shape | Axis-0 | Axis-1 | Axis-2 |
| 1D | (4,) | elements | — | — |
| 2D | (3,3) | rows ↓ | columns → | — |
| 3D | (2,2,2) | depth | rows ↓ | columns → |
17. What is the shape and reshape in NumPy?
Answer: In NumPy, the shape attribute tells you the dimensions of an array (number of rows, columns, depth, etc.). It returns a tuple of integers → one for each axis. Examples of shape import numpy as np # 1D Array arr1 = np.array([1, 2, 3, 4, 5]) print(arr1.shape) ✅ Output: (5,) Meaning: 5 elements in one row (1D array). shape → tells you the size of the array along each axis. reshape() → changes the dimensions (rows, cols, depth) of the array but keeps data the same.18. Explain groupby() in Pandas.
Answer: In Pandas, groupby() is used to split data into groups based on one or more columns, and then apply an aggregation, transformation, or function on each group. It follows the concept of Split → Apply → Combine:- Split the data into groups.
- Apply a function (sum, mean, count, etc.).
- Combine the results into a new DataFrame or Series.
- column_name → the column to group by.
- function() → aggregation (like sum(), mean(), count(), etc.).
19. What is correlation?
Answer: Correlation is a statistical measure that shows the relationship between two variables – how strongly they move together.- It tells us whether an increase in one variable leads to an increase or decrease in another.
- Positive Correlation:
- Both variables move in the same direction.
- Example: Height and weight (taller people usually weigh more).
- Negative Correlation:
- Variables move in opposite directions.
- Example: Speed of a car and travel time (higher speed → lower time).
- Zero (No) Correlation:
- No relationship between variables.
- Example: Shoe size and intelligence.
- The most common measure is the Pearson Correlation Coefficient (r).
- Values range between -1 and +1:
- +1 → Perfect positive correlation.
- -1 → Perfect negative correlation.
- 0 → No correlation.
20. What is a pipeline in Python ML?
Answer: In Machine Learning (ML), a Pipeline is a way to automate the workflow of data preprocessing and model training. It allows you to combine multiple steps (like scaling, encoding, feature selection, training) into a single object, so you don’t have to repeat the same code again and again. Why Do We Use Pipelines?- Keeps code clean and organized.
- Prevents data leakage (ensures preprocessing is only fit on training data).
- Makes experimentation easier (swap one step without rewriting everything).
- Useful in cross-validation (all steps run correctly in each fold).
21. How to handle outliers in Pandas?
Answer: An outlier is a data point that is significantly different from other values in a dataset.- Outliers can occur due to errors (wrong data entry, measurement mistakes) or they may be genuine rare values (like very high salaries in a company dataset).
- Outliers can distort averages (mean).
- They can affect machine learning models (especially linear regression).
- Sometimes they carry important information (e.g., fraud detection).
22. What is the difference between NumPy array and Pandas DataFrame?
Answer: NumPy Array: A multi-dimensional, homogeneous data structure from the NumPy library.- Homogeneous means all elements must be of the same data type (all integers, all floats, etc.).
- It is mainly used for numerical computations in Machine Learning, Data Science, and AI.
- Heterogeneous means it can store different data types in different columns (e.g., integers, strings, floats).
- It is mainly used for data analysis and manipulation (like working with tabular data from Excel, SQL, or CSV).
| Feature | NumPy Array 🚀 | Pandas DataFrame 📊 |
| Library | NumPy | Pandas |
| Data Type | Homogeneous (all values must be same type, e.g., all int or float) | Heterogeneous (different columns can have different data types) |
| Structure | Multi-dimensional array (like a matrix) | Tabular structure with rows & columns |
| Indexing | By integer position only | By labels (row names, column names) and also integers |
| Flexibility | Best for mathematical/numerical operations | Best for handling real-world datasets (CSV, Excel, SQL) |
| Performance | Faster for pure numeric computation | Slightly slower (adds flexibility, labeling, metadata) |
| Use Case | Linear algebra, scientific computing, ML models | Data cleaning, analysis, preprocessing |
23. How to read/write CSV using Pandas?
Answer: Reading a CSV File with Pandas Pandas provides the function pd.read_csv() to read a CSV file into a DataFrame. Example 1: Basic Reading import pandas as pd # Reading CSV file df = pd.read_csv(“employees.csv”) print(df) 👉 Suppose employees.csv contains: Name,Age,Salary Alice,25,50000 Bob,30,60000 Charlie,35,70000 👉 Output: Name Age Salary 0 Alice 25 50000 1 Bob 30 60000 2 Charlie 35 70000 Writing a CSV File with Pandas Pandas provides the function to_csv() to save a DataFrame to a CSV file. Example 1: Basic Writing # Save DataFrame to CSV df.to_csv(“output.csv”, index=False) 👉 This saves: Name,Age,Salary Alice,25,50000 Bob,30,60000 Charlie,35,70000- index=False prevents Pandas from writing row numbers.
24. What is the use of NumPy random module?
Answer:- The NumPy random module is a part of NumPy used to generate random numbers.
- These numbers can be integers, floats, arrays, or samples from probability distributions (like normal distribution, binomial distribution, etc.).
- It is widely used in AI/ML, simulations, testing, and data science.
- To generate synthetic data for testing models.
- To initialize weights randomly in machine learning models.
- To simulate real-world randomness (like rolling dice, flipping coins).
- To sample data from distributions (normal, uniform, binomial, etc.).
25. How to evaluate a dataset’s quality?
Answer: Evaluating dataset quality means checking how reliable, accurate, and useful the dataset is for solving a specific problem. It ensures that the data is clean, consistent, unbiased, and representative of the real-world scenario. Key Factors to Check Dataset Quality- Accuracy
- Data should correctly represent real-world values.
- Example: If a dataset of patients shows a person’s age as 250, that’s inaccurate.
- Completeness
- Check if important values are missing.
- Example: In an e-commerce dataset, if 40% of the “customer email” field is missing, the dataset is incomplete.
- Consistency
- Data should not contradict itself.
- Example: If one record says “Gender = Male” and another record of the same person says “Gender = Female”, that’s inconsistent.
- Uniqueness (No Duplicates)
- Remove duplicate records.
- Example: The same customer appearing 5 times in a sales dataset can bias analysis.
- Validity
- Data should follow the correct format and rules.
- Example: A phone number should have 10 digits, not 5.
- Example: A date of birth cannot be in the future.