Scribd
Upload
24 views3 pages

Day 6 of 100 Data Science Interview Questions Series!!

The document discusses eigenvalue and eigenvector, lemmatization vs stemming, types of recommendation systems, bias-variance tradeoff, and vanishing/exploding gradients. Eigenvalue and eigenvector can be used to reduce dataset dimensions while maintaining key information. Lemmatization and stemming are text normalization techniques, with lemmatization being better for QnA bots and stemming for sentiment analysis. Common recommendation system types include collaborative, content-based, demographic-based, and knowledge-based systems.

Uploaded by

Silga
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
24 views3 pages

Day 6 of 100 Data Science Interview Questions Series!!

The document discusses eigenvalue and eigenvector, lemmatization vs stemming, types of recommendation systems, bias-variance tradeoff, and vanishing/exploding gradients. Eigenvalue and eigenvector can be used to reduce dataset dimensions while maintaining key information. Lemmatization and stemming are text normalization techniques, with lemmatization being better for QnA bots and stemming for sentiment analysis. Common recommendation system types include collaborative, content-based, demographic-based, and knowledge-based systems.

Uploaded by

Silga
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Day 6 of 100 Data Science Interview Questions Series!!

Q 26.) How can you use eigenvalue or eigenvector?

It ​is​ difficult to understand ​and​ visualize data ​with​ more than ​3


dimensions, let alone a dataset of over ​100​+ dimensions. Hence, it
would be ideal to somehow compress/transform this data into a smaller
dataset. This is where we can use this concept.

We can utilise Eigenvalues ​and​ Eigenvectors to ​reduce​ the dimension


space ensuring most of the key information ​is​ maintained.

Eigenvalues ​are the directions along which a particular linear


transformation acts by flipping, compressing, ​or​ stretching.

Eigenvectors ​are ​for​ understanding linear transformations. In data


analysis, we usually calculate the eigenvectors ​for​ a correlation ​or
covariance matrix.

Please view this article which has explained this concept better than I
ever could!
https://medium.com/fintechexplained/what-are-eigenvalues-and-eigenvecto
rs-a-must-know-concept-for-machine-learning-80d0fd330e47

Q 27.) What is lemmatization and Stemming, Which one should I use in Sentimental
Analysis, and which one should I use in QnA bot?

They are used ​as​ Text Normalization techniques ​in​ NLP ​for
preprocessing text.

Stemming ​is​ the process of reducing inflection ​in​ words to their root
forms such ​as​ mapping a group of words to the same stem even ​if​ the
stem itself ​is​ ​not​ a valid word ​in​ the Language.​"

Lemmatization​, unlike Stemming, reduces the inflected words properly


ensuring that the root word belongs to the language. In Lemmatization
root word ​is​ called Lemma.
● Stemming ​is​ better option ​for​ Sentimental Analysis ​as​ the meaning
of the word ​is​ ​not​ necessary ​for​ understanding sentiments, ​and
stemming ​is​ little faster than Lemmatization.
● Lemmatization ​is​ better ​for​ QnA bot ​as​ word should have a proper
emaning ​while​ conversing ​with​ a human subject.

Q 28.) What are some common Recommendation System Types, where can I use them?

Recommendation systems are used to recommend ​or​ generate some


outputs based on previous inputs that were given by users.
Recommendation system can be build thorugh Deep Learning, like Deep
Beliefe networks, RBM,AutoEncoder etc ​or​ some traditional techniques.

Some common types are:

1. Collaborative Recommender system


2. Content-based recommender system
3. Demographic based recommender system
4. Utility based recommender system
5. Knowledge based recommender system
6. Hybrid recommender system.

● DL based Recommendation systems can be used ​for​ dimentionality


reduction ​and​ generating similar output.
● RS can also be used ​for​ suggestions of similar items based on
user​'s past choices and item'​s content.
● RS can also be used ​for​ suggestions of simlar products based on a
group of users ​with​ simlar features ​as​ you.

Q 29.) What is bias, variance trade-off?

​ s​ error introduced ​in​ your model due to over simplification


Bias i
of machine learning algorithm.” It can lead to under fitting.

● Low bias machine learning algorithms — Decision Trees, k-NN ​and


SVM
● High bias machine learning algorithms — Linear Regression,
Logistic Regression
Variance ​is​ error introduced ​in​ your model due to ​complex​ machine
learning algorithm, your model learns noise also ​from​ the training data
set​ ​and​ performs bad on test data ​set​. It can lead high sensitivity ​and
over fitting.

Normally, ​as​ you increase the complexity of your model, you will see a
reduction ​in​ error due to lower bias ​in​ the model. However, this only
happens till a particular point​. As you ​continue​ to make your model
more ​complex​, you end up over-fitting your model ​and​ hence your model
will start suffering ​from​ high variance.​ ​Increasing the bias will
decrease the variance. Increasing the variance will decrease the bias.
This is Bias-Variance Trade-Off.

Q 30.) What are vanishing/exploding gradients?

Gradient ​is​ the direction ​and​ magnitude calculated during


training of a neural network that ​is​ used to update the network weights
in​ the right direction ​and​ by the right amount.

● Exploding gradient ​is​ a problem where large error gradients


accumulate ​and​ result ​in​ very large updates to neural network
model weights during training.
● Vanishing gradient ​is​ a problem where ​as​ more layers are added to
neural networks, the gradients of the loss function approaches
zero, making the network hard to train. This occurs ​in​ large
models ​with​ many layers. ​Models like ResNet, that have skip
connections, are a good solution to this problem.

- Alaap Dhall

Follow ​Alaap Dhall​ on LinkedIn for more insights in Data


Science and Deep Learning!!

Visit ​https://www.aiunquote.com​ for a 100 project series in


Deep Learning.

You might also like