26/07/2020 Introduction to Data Analytics - - Unit 9 - Week 8 - Clustering Analysis and Prescriptive Analytics

reviewer2@nptel.iitm.ac.in ▼

Courses » Introduction to Data Analytics Announcements Course Ask a Question Progress

Unit 9 - Week 8 - Clustering Analysis and Prescriptive Analytics

Course outline Assignment 8

How to access the portal The due date for submitting this assignment has passed. Due on 2017-09-20, 23:55 IST.
As per our records you have not submitted this assignment.
Week 1 - Course
Overview and Descriptive 1) In solving a n-armed bandit problem, we decide to use the -greedy algorithm. We have the following two choices for the 1 point
Statistics 𝜖 parameter, 𝜖 = 0.1 and 𝜖 = 0.01 . If we run the algorithm for a very long time, and want to maximize the cumulative reward, which value of
𝜖 should we use?

Week 2 - Probability Suppose we want to find an optimal arm in the fastest time. In this situation, which value of 𝜖 should we use?
Distributions & Inferential
Statistics

Week 3 - Inferential for maximum reward use 𝜖 = 0.1 ; for minimising the time required to find an optimal
Statistics arm, use 𝜖 = 0.01

Week 4 - Machine
for maximum reward use 𝜖 = 0.01 ; for minimizing the time required to find an optimal arm, use 𝜖 = 0.1
Learning

Week 5 - Supervised for both cases, use 𝜖 = 0.1

Learning (Regression and

Classification for both cases, use 𝜖 = 0.01
Techniques) - I
No, the answer is incorrect.
Week 6 : Supervised Score: 0
Learning (Regression and
Classification Accepted Answers:
Techniques)-II for maximum reward use 𝜖 = 0.01 ; for minimizing the time required to find an optimal arm, use 𝜖 = 0.1

2) Suppose that you have been given a number of different drug formulations to treat a particular disease and your job is to identify one 1 point
Week 7 - Association
Rule Mining and Big Data among them that best meets certain criteria with regards to its efficacy in treating the disease. Before you run the experiments, you need to
provision for the samples that would be required. Treating this as a n-armed bandit problem, which kind of solution method would you prefer for
Week 8 - Clustering identifying the best option?
Analysis and Prescriptive
Analytics asymptotic correctness
regret optimality
Clustering Analysis
PAC optimality
Clustering Analysis
(cont'd) No, the answer is incorrect.
Score: 0
Introduction to
Experimentation and Accepted Answers:
Active Learning PAC optimality

Introduction to 3) After 12 iterations of the UCB algorithm applied on a 4-arm bandit problem, we have 2 points
Experimentation and 𝑛1 = 3, 𝑛2 = 4, 𝑛3 = 3, 𝑛4 = 2and𝑥¯1 = 0.55, 𝑥¯2 = 0.63, 𝑥¯3 = 0.61, 𝑥¯4 = 0.40 . Which arm should be played next?
Active Learning(cont'd)
1
An Introduction to Online
Learning - Reinforcement 2
Learning 3
An Introduction to Online 4
Learning - Reinforcement
Learning (cont'd) No, the answer is incorrect.
Score: 0
Quiz : Assignment 8
Accepted Answers:
Feedback for week 8 4
Assignment 8: Solution

Course Summary+
Insight into the Final
Exam

https://onlinecourses-archive.nptel.ac.in/noc17_mg24/unit?unit=62&assessment=98 1/4
26/07/2020 Introduction to Data Analytics - - Unit 9 - Week 8 - Clustering Analysis and Prescriptive Analytics
4) Consider the following image showing data points belonging to three different clusters (indicated by the colors of the points). 1 point

If we run the k-means clustering algorithm with k = 3, do you think the algorithm will be able to correctly cluster the data points belonging to the
three clusters?

no
yes

No, the answer is incorrect.

Score: 0
Accepted Answers:
no

5) Consider the following one dimensional data set: 12, 22, 2, 3, 33, 27, 5, 16, 6, 31, 20, 37, 8, 18. Given k = 3 and initial cluster 1 point
centers to be 5, 6 and 31, what are the final cluster centers obtained on applying the k-means algorithm?

5, 18, 30
5, 18, 32
6, 19, 32
4.8, 17.6, 32

No, the answer is incorrect.

Score: 0
Accepted Answers:
4.8, 17.6, 32

6) In outlier mining, we try to detect data points that do not fit a particular distribution. Which among the following are disadvantages of 1 point
using DBSCAN for performing outlier mining?
(Note that more than one statement may be correct)

No need to make any changes to the DBSCAN algorithm.

DBSCAN does not work well when there is large variation in the density of the data point groupings.
DBSCAN can detect outlier points which lie between groupings of data points and not just points far away from all other data points.
Slight change in the DBSCAN parameters can lead to large variation in the resultant outlier points detected.

No, the answer is incorrect.

Score: 0
Accepted Answers:
DBSCAN does not work well when there is large variation in the density of the data point groupings.
Slight change in the DBSCAN parameters can lead to large variation in the resultant outlier points detected.

7) What would, in general, be the effect of increasing MinPts in DBSCAN while retaining the same Eps parameter? (Note that more 1 point
than one statement may be correct)

Increase in the sizes of individual clusters

Decrease in the sizes of individual clusters
Increase in the number of clusters
Decrease in the number of clusters

No, the answer is incorrect.

Score: 0
Accepted Answers:
Decrease in the sizes of individual clusters
Increase in the number of clusters

8) Given a two-class training data set with 100 unlabelled data points, suppose we randomly select 10 data points and query for their 1 point
labels. We supply these 10 labelled data points to a SVM, and obtain a decision boundary. Assuming a limit on the number of additional points
that we can select to improve this classifier, in general, would you prefer to query the labels of points lying close to the decision surface or those
that are far from the decision surface?

close to the decision surface

https://onlinecourses-archive.nptel.ac.in/noc17_mg24/unit?unit=62&assessment=98 2/4
26/07/2020 Introduction to Data Analytics - - Unit 9 - Week 8 - Clustering Analysis and Prescriptive Analytics
far from the decision surface

No, the answer is incorrect.

Score: 0
Accepted Answers:
close to the decision surface

9) For the similarity matrix given below, show the hierarchy of clusters created by the single link clustering algorithm. 1 point

a
b
c
d

No, the answer is incorrect.

Score: 0
Accepted Answers:
b

10) For the similarity matrix given in the previous question, show the hierarchy of clusters created by the complete link clustering 1 point
algorithm.

https://onlinecourses-archive.nptel.ac.in/noc17_mg24/unit?unit=62&assessment=98 3/4
26/07/2020 Introduction to Data Analytics - - Unit 9 - Week 8 - Clustering Analysis and Prescriptive Analytics

a
b
c
d

No, the answer is incorrect.

Score: 0
Accepted Answers:
d

Previous Page End

© 2014 NPTEL - Privacy & Terms - Honor Code - FAQs -

A project of In association with Funded by Powered by

https://onlinecourses-archive.nptel.ac.in/noc17_mg24/unit?unit=62&assessment=98 4/4