uiz Space

September 2023 term · Linear Statistical Models · BSMA3012

Linear Statistical Models Quiz 1: 29 October 2023 (September 2023 term)

The IIT Madras BS Linear Statistical Models (Linear Statistical Models) Quiz 1 paper sat on 29 Oct 2023, in the September 2023 term: 1 question for 30 marks in 120 minutes. Every question is below with its answer. Take it as a timed mock test to be marked, or read it through first.

Questions
1
Marks
30
Duration
120 min
MCQ
1

Updated

Official paper: IIT M DEGREE AN2 EXAM QPE2 29 Oct 2023 · No negative marking.

Question 1

+30 marksOne correct option
  1. Let xx and yy be vectors of integers.

(i) Write an R code to check if the given two vectors, say xx and yy, have the same length or not. [2 Marks]

(ii) Write an R code to compute the dot product of two numeric vectors xx and yy. [2 Marks]

(iii) Write an R code to create a new vector containing only the even numbers from the vector xx. [3 Marks]

  1. Suppose Y⃗=Xβ⃗+ϵ⃗\vec{Y} = X\vec{\beta} + \vec{\epsilon} be a linear model, where

Y⃗=(y1,y2,y3,…,yn)T,β⃗=(β1,β2,β3,…,βm)T,X=((xij))  ;1≤i≤n,1≤j≤m\begin{aligned} \vec{Y} &= (y_1, y_2, y_3, \ldots, y_n)^T, \\ \vec{\beta} &= (\beta_1, \beta_2, \beta_3, \ldots, \beta_m)^T, \\ X &= ((x_{ij})) \;; 1 \leq i \leq n, 1 \leq j \leq m \end{aligned}

and

ϵ⃗=(ϵ1,ϵ2,ϵ3,…,ϵn)T\vec{\epsilon} = (\epsilon_1, \epsilon_2, \epsilon_3, \ldots, \epsilon_n)^T

where, ϵ⃗\vec{\epsilon} has mean 0⃗\vec{0} and variance covariance matrix σ2In×n\sigma^2 I_{n \times n}.

(a) Define what is meant by β⃗^\hat{\vec{\beta}} is the least square estimate of β⃗\vec{\beta}? [3 Marks]

(b) Show that any least square estimate β⃗^\hat{\vec{\beta}} of β⃗\vec{\beta} satisfies:

(XTX)β⃗^=XTY⃗(X^T X)\hat{\vec{\beta}} = X^T \vec{Y}

[5 Marks]

  1. An analyst wanted to predict the number of goals scored by a Canadian football club, for which he/she collected the following data:
Temperature on match day (x⃗\vec{x})Number of Goals (y⃗\vec{y})
-14
01
13
20

(a) Draw a scatter plot of (x⃗,y⃗)(\vec{x}, \vec{y}). [2 Marks]

(b) Let

X=(1−1101112)X = \begin{pmatrix} 1 & -1 \\ 1 & 0 \\ 1 & 1 \\ 1 & 2 \end{pmatrix}

Suppose we assume the linear model (y⃗,Xβ⃗,I5×5)(\vec{y}, X\vec{\beta}, I_{5\times5}) for the data. Find the least square estimate for β⃗\vec{\beta}. [4 Marks]

(c) For the values of β⃗\vec{\beta} obtained, draw the least square line on the graph above. [2 Marks]

(d) Another analyst claim that the least square estimate for β⃗\vec{\beta} is given by:

β⃗~=β⃗^+(1−1)\tilde{\vec{\beta}} = \hat{\vec{\beta}} + \begin{pmatrix} 1 \\ -1 \end{pmatrix}

where, β⃗^\hat{\vec{\beta}} is the least square estimate obtained in (c).

Whose claim for least square estimate for β⃗\vec{\beta} is correct and why? Elaborate [4Marks]

(e) Estimate the value of yy when x=5,−2x = 5, -2. [3 Marks]

  1. A

    I have written answers on the answer sheets

  2. B

    Not applicable

Show answer

Correct answer

  • A

    I have written answers on the answer sheets