0% found this document useful (0 votes)
101 views

Solution Key Comprehensive Question Paper

This document provides instructions for a mathematical foundations for data science comprehensive examination at Birla Institute of Technology & Science, Pilani. It includes 12 questions worth 5 marks each on topics related to linear algebra, data analysis, and optimization. Students are instructed to attempt any 8 questions, showing their work and assumptions. Sample solutions are provided for some of the questions as examples.

Uploaded by

ADITYA K R
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
0% found this document useful (0 votes)
101 views

Solution Key Comprehensive Question Paper

This document provides instructions for a mathematical foundations for data science comprehensive examination at Birla Institute of Technology & Science, Pilani. It includes 12 questions worth 5 marks each on topics related to linear algebra, data analysis, and optimization. Students are instructed to attempt any 8 questions, showing their work and assumptions. Sample solutions are provided for some of the questions as examples.

Uploaded by

ADITYA K R
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/ 8

Birla Institute of Technology & Science, Pilani

Work Integrated Learning Programmes Division


Second Semester 2022-2023

Solution Key-Comprehensive Examination


(EC-3 Regular)

Course No. : SSCSHZC416


Course Title : Mathematical Foundations for Data Science
Nature of Exam : Open Book
Weightage : 40%
Duration : 2 Hours
Date of Exam : 02-09-23
Note to Students:
1. Please follow all the Instructions to Candidates given on the cover page of the answer book.
2. All parts of a question should be answered consecutively. Each answer should start from a fresh page.
3. Assumptions made if any, should be stated clearly at the beginning of your answer.
4. Attempt any 8 questions.

4 10 8 � 0 0 � � �
Q.1. If 10 26 26 = � � 0 0 � � find the values of a,b,c,d,e,f and show the
8 26 61 � � � 0 0 �
calculations in details too. [5]
Solution :
4 10 8 � 0 0 � � �
10 26 26 = � � 0 0 � �
8 26 61 � � � 0 0 �

2 marks for matrix Multiplication 3 marks for calculating a,b,c,d,e,f (0.5 each)
and just because of "c" and "d" position is interchanged in 2nd matrix (typo error), so i am giving
full marks to you.

Q.2. Find the largest eigenvalue and the corresponding eigenvector of the matrix
10 7 8 7
7 5 6 5 [5]
8 6 10 9
7 5 9 10
Solution:
So largest eigenvalue is 30.2887
And Corresponding eigenvector is (1.01467, 0.729975, 1.05957, 1).
1mark for ch equation
2 marks for largest eigenvalue
2 marks for largest eigenvector

Q.3. Obtain the transformation from �1 , �2, �3 to �1 , �2 , �3 when


�1 = 2�1 + �2 , �2 = �2 − 2�3 , �3 =− �1 + 2�2 + �3 and �1 = �1 + �2 + �3 ,
�2 = �1 + 2�2 + 3�3 , �3 = �1 + 3�2 + 5�3 , [5]
Solution :
2 1 0
The matrix of the transformation from x to y is A= 0 1 −2
−1 2 1
1 1 1
The matrix of the transformation from y to z is B= 1 2 3
1 3 5
Hence, The matrix of the transformation from x to z is
1 1 1 2 1 0
C = BA = 1 2 3 0 1 −2
1 3 5 −1 2 1
2+0−1 1+1+2 0−2+1 1 4 −1
= 2 + 0 − 3 1 + 2 + 6 0 − 4 + 3 = −1 9 −1
2 + 0 − 5 1 + 3 + 10 0 − 6 + 5 −3 14 −1
Therefore,
�� = �� + ��� − ��
�� =− �� + ��� − ��
�� = −��� + ���� − ��

2 Marks for Calculation of intermediate steps and 3 marks for �� , �� and ��

Q.4. Prove that a linear transformation T: V → W is non-singular ⟺ T carries a linearly


independent Subset of V to a linearly independent Subset of W. [5]

Solution :

2.5 marks for one way and 2.5 marks for the other way

Q.5. Let (x1, y1),(x2, y2), . . .(xN , yN ) be N points on which we find a SVM classifier using the dual
SVM formulation which is given below:

Maximize
� � �

�� − �� �� �� �� ����
�=1 �=1 �=1
Subject to

�� �� = 0 ��� �� ≥ 0, ∀�
�=1
Let OA be the value of the objective function at the optimal solution returned by the dual SVM
formulation for this problem. Now we add a new point (xN+1, yN+1) and find a SVM classifier by
solving the dual formulation again. Let OB be the value of the objective function at the optimal
solution for this problem. Considering the following three relationships
(a) OA > OB,
(b) OA = OB,
(c) OA < OB
Determine which of these relationships are possible and give a mathematical argument for your
answer in each case. [5]
Solution :
Q.6. Let x1, x2, . . . xN be N points on which we perform Principle Components Analysis leading to
the discovery of the principle component directions b1, b2 . . . bD. The given data is now
transformed to the points y1, y2, . . . yN where yi = Qxi + µ where QTQ = I and µ is a constant
vector. Determine the principle components for the transformed set of points in terms of the
old principle components. How much variance is accounted for by the first principal
component for the transformed set of points in terms of the variance accounted for by the
first principle component for the original set of points? Justify your answer with
mathematical arguments. [5]
Solution :

Q.7. Consider three linearly independent vectors in Rn named a1, a2 and a3. Now construct three
vectors b1 = a2 − a3, b2 = a1 − a3 and b3 = a1 − a2. Now consider the set Q = {b1, b2, b3}.
Prove or disprove that the set Q is linearly independent. [5]
Solution :

Q.8. Consider two sets named H1 and H2. It is known that these two sets are convex sets.
(a) Prove or disprove that H1 ∩ H2 is a convex set. Here ∩ represents the set intersection
operation.
(b) Prove or disprove that H1 ∪ H2 is a convex set. Here ∪ represents the set union
operation. [5]
Solution :
Q.9. A linear Algebra student arrived at an n × n real matrix given below
�11 �12 �1�
�21 �22 ⋯ �2�
A= ⋮ ⋱ ⋮
��1 ��2 ⋯ ���

Help the student to find singular value decomposition of full rank matrix A if the columns of
A are orthogonal. [5]
Solution :
Q.10. A data analyst modeled the objective function f : Rn → R as product of squares of n feature
xi’s and n ≥ 2. He has to maximize the objective function such that sum of squares of n
features is less than or equal to c2 where c ∈ R. Write the mathematical formulation of the
problem and solve it. Using the above result prove the inequality

(a1a2 · · · an)1/n ≤ (a1 + · · · + an)/n for any ai > 0, i = 1, · · · , n [5]


Solution :
Q.11. You are given the quadratic polynomial
f(x, y, z) = 2x2 − 2xy − 4xz + y2 + 2yz + 3z2 − 2x + 2z:
(a) Write f(x, y, z) in the form f(x, y, z) = xTAx − bT x where x = (x, y, z), A is a real symmetric matrix,
and b is constant vector.
(b) Find the point (x, y, z) where f(x, y, z) is at an extremum.
(c) Is this point a minimum, maximum, or a saddle point of some kind? [5]
Solution :
Q.12. Solve the System of equations by Gaussian elimination method
2x1 + 5x2 + 2x3 − 3x4 = 3
3x1 + 6x2 + 5x3 + 2x4 = 2
4x1 + 5x2 + 14x3 + 14x4 = 11
5x1 + 10x2 + 8x3 + 4x4 = 4 [5]
Solution :

***********

You might also like