Q$1.$ Set A

Explain what is margin is Support Vector Machine $($SVM$)$ algorithm? Explain what are Support Vectors? $(5$ marks$)$

Q$1.$ Set B

What do you understand by Kernel Trick with respect to Support Vector Machine $($SVM$)$ algorithm? Explain what are Support Vectors? $(5$ marks$)$

Q$1.$ Set C

Is Support Vector Machine $($SVM$)$ a parametric machine learning technique or a non$-$parametric technique? Explain. What are support vectors in SVM$?(5$ marks$)$

Q$2.$ Set A

What is the main assumption of Naive Bayes classifier over Bayes Theorem? How does it simplify the modelling process? $(5$ Marks$)$

Q$2.$ Set B

Explain what is Bayes Theorem, how is it related to the Naive Bayes classifier? $(5$ Marks$)$

Q$2.$ Set C

How do we deal with continuous variables in Nave Bayes Classifier? Explain. $(5$ Marks$)$

Q$3.$ Set A

How does Border point differ from Core Point in DBScan algorithm? Explain. $(5$ Marks$)$

Q$3.$ Set B

In what situation is DBScan advantageous to use than K$-$Means Clustering algorithm? Explain. $(5$ Marks$)$

Q$3.$ Set C

What will happen if MinPts $($Minimum number of points$)$ parameter is chosen too small or large in DBScan algorithm? Explain. What are the different parameters in DBSCAN algorithm? $(5$ marks$)$

Q$4$ Set A$.$

Explain what is the drawback of KMeans algorithm with respect to initialization of centroids? How can it be resolved? Explain the steps. $(10$ Marks$)$

Q$4.$ Set B$.$

Suppose you have a dataset of customer transactions for a retail store, consisting of customer IDs, purchase items, and purchase quantities. How can hierarchical $($agglomerative$)$ clustering be applied to analyze customer purchasing patterns and segment customers based on their buying behavior? Explain the steps of the entire pipeline. $(10$ Marks$)$

Q$4.$ Set C$.$

In the context of e$-$commerce, how can the K$-$means$++$ algorithm be applied to perform customer segmentation and improve targeted marketing strategies? Explain what is the drawback of KMeans algorithm that is solved by K$-$means$++$ in this case. $(10$ Marks$)$

Q$5.$ Set A$.$

Let's consider a use case where PCA is applied to a dataset containing various features related to customer behavior in an e$-$commerce setting. The dataset includes variables such as purchase frequency, average order value, time spent on the website, and number of products viewed. Following are the loading vectors generated for PC$1$ and PC$2$ respectively. $(10$ marks$)$

V$1$

V$2$

purchase frequency

$0\mathrm{.}7$

$-0\mathrm{.}2$

average order value

$0\mathrm{.}2$

$0\mathrm{.}6$

time spent on the website

$0\mathrm{.}5$

$-0\mathrm{.}3$

number of products

$0\mathrm{.}6$

$0\mathrm{.}7$

Interpret the loading vectors with respect to PC$1$ and PC$2.$ Comment on what does a datapoint with high PC$1$ and a datapoint with high PC$2$ indicate regarding the customer segment it belongs to$.$

Q$5.$ Set B

Let's consider a use case where PCA is applied to a dataset containing various features related to customer behavior in an e$-$commerce setting. The dataset includes variables such as purchase frequency, average order value, time spent on the website, and number of products viewed. Following are the loading vectors generated for PC$1$ and PC$2$ respectively. $(10$ marks$)$

V$1$

V$2$

purchase frequency

$0\mathrm{.}7$

$0\mathrm{.}8$

average order value

$0\mathrm{.}2$

$-0\mathrm{.}1$

time spent on the website

$0\mathrm{.}5$

$-0\mathrm{.}4$

number of products

$0\mathrm{.}6$

$0\mathrm{.}6$

Interpret the loading vectors with respect to PC$1$ and PC$2.$ Comment on what does a datapoint with high PC$1$ and a datapoint with high PC$2$ indicate regarding the customer segment it belongs to$.$

Q$5.$ Set C

Let's consider a use case where PCA is applied to a dataset containing various features related to customer behavior in an e$-$commerce setting. The dataset includes variables such as purchase frequency, average order value, time spent on the website, and number of products viewed. Following are the loading vectors generated for PC$1$ and PC$2$ respectively. $(10$ marks$)$

V$1$

V$2$

purchase frequency

$-0\mathrm{.}7$

$-0\mathrm{.}2$

average order value

$-0\mathrm{.}2$

$0\mathrm{.}6$

time spent on the website

$-0\mathrm{.}5$

$-0\mathrm{.}3$

number of products

$-0\mathrm{.}6$

$0\mathrm{.}7$

Interpret the loading vectors with respect to PC$1$ and PC$2.$ Comment on what does a datapoint with high PC$1$ and a datapoint with high PC$2$ indicate regarding the customer segment it belongs to$.$

Q$6.$ Set A$.$

Which activation function can be used in the output layers of two different Artificial Neural Network designed for the following two use cases. Explain how the corresponding activation functions work and why are they suitable for their corresponding use cases. $(10$ Marks$)$

A healthcare organization wants to develop a system to diagnose diseases based on a set of medical symptoms provided by patients. The goal is to classify the diseases into mutually exclusive categories, such as "Common Cold," "Influenza," and "Strep Throat," based on the symptoms reported by the patients.

A movie streaming platform wants to classify movies into multiple genres based on their plot summaries and metadata. The goal is to categorize each movie into multiple non$-$exclusive genres, such as "Action," "Comedy," "Drama," "Romance," and "Thriller," to improve movie recommendations and user experience.

Q$6.$ Set B$.$

Which activation function can be used in the output layers of two different Artificial Neural Network designed for the following two use cases. Explain how the corresponding activation functions work and why are they suitable for their corresponding use cases. $(10$ Marks$)$

A healthcare organization wants to develop a system to diagnose diseases based on a set of medical symptoms provided by patients. The goal is to classify the diseases into mutually exclusive categories, such as "Common Cold," "Influenza," and "Strep Throat," based on the symptoms reported by the patients.

A movie streaming platform wants to classify movies into multiple genres based on their plot summaries and metadata. The goal is to categorize each movie into multiple non$-$exclusive genres, such as "Action," "Comedy," "Drama," "Romance," and "Thriller," to improve movie recommendations and user experience.

Q$6.$ Set C$.$

Which activation function can be used in the output layers of two different Artificial Neural Network designed for the following two use cases. Explain how the corresponding activation functions work and why are they suitable for their corresponding use cases. $(10$ Marks$)$

A healthcare organization wants to develop a system to diagnose diseases based on a set of medical symptoms provided by patients. The goal is to classify the diseases into mutually exclusive categories, such as "Common Cold," "Influenza," and "Strep Throat," based on the symptoms reported by the patients.

A movie streaming platform wants to classify movies into multiple genres based on their plot summaries and metadata. The goal is to categorize each movie into multiple non$-$exclusive genres, such as "Action," "Comedy," "Drama," "Romance," and "Thriller," to improve movie recommendations and user experience.