Habib Shaikh
AI Expert
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
What is Data Science?
Data Science is a multidisciplinary field that
blends various techniques to analyze large data
sets and derive actionable insights.
Key Components
Data Collection: Gathering raw data from various
sources.
Data Cleaning: Removing errors and
inconsistencies for accuracy.
Data Storage: Warehousing and structuring data
for accessibility.
Analysis Methods: Applying statistical,
mathematical, and machine learning algorithms.
Visualization: Presenting findings through charts
and graphs for better understanding.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
Define the terms KPI, lift, model
fitting, robustness, and DOE.
KPI (Key Performance Indicator): A measure of
how effectively a business is achieving its
objectives.
Lift: A metric used to compare the performance
of a target model against a random choice model.
Lift quantifies how well the model predicts
compared to no model at all.
Model Fitting: Refers to how accurately a model
corresponds to the given data observations.
Robustness: The ability of a system to handle
variability or noise effectively.
DOE (Design of Experiments): A structured
approach to investigating and explaining the
variation of information under assumed
conditions by reflecting variables.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
What is the difference between
data analytics and data science?
Data science involves transforming data using
advanced analysis methods to extract insights,
which can then be applied in various business
contexts. In contrast, data analytics focuses on
examining existing data to validate hypotheses
and support decision-making.
Data science is forward-looking, involving
predictive modeling and innovations for future
problem-solving, while data analytics is more
focused on understanding past trends for
immediate business decisions. Data science
encompasses a wider scope of techniques, tools,
and methodologies, while data analytics typically
deals with more specific, concentrated issues.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
What are some sampling
techniques and the advantages
of sampling?
Due to the challenges of analyzing large datasets
in their entirety, sampling allows for the selection
of representative data points for analysis. Two
main categories of sampling techniques are:
Probability Sampling: Techniques like cluster
sampling, simple random sampling, and stratified
sampling.
Non-Probability Sampling: Methods such as
quota sampling, convenience sampling, and
snowball sampling.
Sampling ensures that analysis remains
manageable while still representing the whole
dataset.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
List the conditions for overfitting
and underfitting.
Overfitting: Occurs when a model works well on
training data but fails on new data. This happens
due to low bias and high variance, often seen in
decision trees.
Underfitting: Happens when a model is too
simple to capture the underlying data
relationships, leading to poor performance even
on training data. It results from high bias and low
variance, typical of linear regression.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
Differentiate between long and
wide format data.
Long Format Data: Each row represents one data
point per subject, with multiple rows for each
subject. Common in R analysis and data logging.
Wide Format Data: Multiple observations per
subject are stored in separate columns. This
format is typically used for repeated measures
ANOVA in statistical packages.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
What are Eigenvectors and
Eigenvalues?
Eigenvectors are unit vectors whose magnitude
equals one and are used in eigen decomposition
of matrices.
Eigenvalues are coefficients applied to these
vectors, altering their magnitude. These
concepts are essential in machine learning
techniques like PCA, where they help identify
significant patterns in data.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
What do high and low p-values
mean?
A p-value indicates the likelihood that the
observed results are due to chance.
Low p-value (≤ 0.05): Suggests that the null
hypothesis can be rejected, indicating the result
is statistically significant.
High p-value (≥ 0.05): Indicates the null
hypothesis remains valid, suggesting the
observed data is likely due to random variation.
p-value of 0.05: Implies a borderline result
where the hypothesis could go either way.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
When is resampling done?
Resampling is employed to assess the stability
and accuracy of a model, typically by training the
model on various subsets of the data. It helps
quantify uncertainties, ensuring the model can
handle diverse patterns in the data and is
validated through different random selections.
What is imbalanced data?
Imbalanced data refers to a dataset where
certain categories or classes are
underrepresented, leading to biases in model
predictions and accuracy issues.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
Are there differences between
expected value and mean value?
While both represent central tendencies, the
expected value pertains to random variables,
whereas the mean is related to probability
distributions. The expected value is often used in
stochastic processes, while the mean is a
general statistic for averaged data.
What is Survivorship Bias?
Survivorship bias is the error made when
focusing on successful subjects and ignoring
those that failed, leading to false conclusions
based on incomplete data. This can distort
analyses and create misleading interpretations.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
What is Gradient and Gradient
Descent?
Gradient: A vector indicating how much the
output of a function changes relative to changes
in its input. It represents the slope of the
function.
Gradient Descent: A technique used to minimize
a function by iteratively adjusting the input in the
direction of the steepest decrease, often applied
to minimize loss functions in machine learning.
Define confounding variables.
Confounding variables are extraneous factors
that affect both the independent and dependent
variables, causing spurious relationships that can
distort the conclusions drawn from data analysis.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
Explain the bias-variance trade-
off.
Bias: Error introduced by oversimplifying the
model. Low-bias models like decision trees are
complex, while high-bias models like linear
regression are simpler.
Variance: Error due to model complexity, where
overly complex models may overfit and perform
poorly on unseen data.
The trade-off suggests that as model complexity
increases, bias decreases, but variance may
increase, leading to overfitting. The goal is to
find an optimal balance for model accuracy.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
Define the confusion matrix.
A confusion matrix is a 2x2 matrix used in
classification tasks to evaluate a model's
performance. It shows the number of correct and
incorrect predictions, broken down into four
categories: True Positive (TP), False Positive
(FP), True Negative (TN), and False Negative
(FN). From these, metrics like accuracy,
precision, recall, and F-score are derived.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
What is logistic regression?
Provide an example of its use.
Logistic regression is a technique used to model
binary outcomes using a linear combination of
predictor variables. For example, predicting
election outcomes based on factors like
campaign spending and political history.
What is Linear Regression and its
drawbacks?
Linear regression models the relationship
between a dependent variable and independent
variables. Drawbacks include assumptions of
linearity, inability to model binary outcomes, and
vulnerability to overfitting.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
What is Random Forest and how
does it work?
Random Forest is an ensemble learning technique
that builds multiple decision trees and combines
their outputs to improve classification accuracy.
Each tree is trained on random subsets of data,
and predictions are made based on the majority
vote across all trees.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
Calculate the chance of seeing a
shooting star within an hour
given a 0.2 probability every 15
minutes.
With a 0.2 chance of seeing a shooting star in 15
minutes, the probability of not seeing any in 15
minutes is 0.8. Over an hour (four 15-minute
intervals), the chance of seeing no stars is 0.8⁴ ≈
0.40. Thus, the chance of seeing at least one star
is 1 - 0.40 = 0.60 or 60%.
DATA SCIENCE Habib Shaikh
AI Expert
INTERVIEW QUESTIONS FOR FRESHERS
What is deep learning and its
difference from machine
learning?
Deep learning is a subset of machine learning
that uses layered neural networks to process and
learn from data. Unlike traditional machine
learning, which uses simpler models, deep
learning simulates the human brain's structure for
higher accuracy and feature extraction.
