linear regression assignment pdf

Shalabh [email protected] [email protected] Department of Mathematics & Statistics Indian Institute of Technology Kanpur , Kanpur - 208016 ( India )

MTH 416 : Regression Analysis

Syllabus: Simple and multiple linear regression, Polynomial regression and orthogonal polynomials, Test of significance and confidence intervals for parameters. Residuals and their analysis for test of departure from the assumptions such as fitness of model, normality, homogeneity of variances, detection of outliers, Influential observations, Power transformation of dependent and independent variables. Problem of multicollinearity, ridge regression and principal component regression, subset selection of explanatory variables, Mallow's Cp statistic. Nonlinear regression, different methods for estimation (Least squares and Maximum likelihood), Asymptotic properties of estimators. Generalised Linear Models (GLIM), Analysis of binary and grouped data using logistic and log-linear models.  

Grading Scheme : Quizzes: 20%, Mid semester exam: 30%, End semester exam: 50%

Books:  1. Introduction to Linear Regression Analysis by Douglas C. Montgomery, Elizabeth A. Peck, G. Geoffrey Vining (Wiley), Low price Indian edition is available.

                2. Applied Regression Analysis by Norman R. Draper, Harry Smith (Wiley), Low price Indian edition is available.

                3. Linear Models and Generalizations - Least Squares and Alternatives by  C.R. Rao, H. Toutenburg, Shalabh, and C. Heumann (Springer, 2008)

                4. A Primer on Linear Models by John F. Monahan (CRC Press, 2008)

                5. Linear Model Methodology by Andre I. Khuri (CRC Press, 2010)

Assignaments:

Assignment 1

Assignment 2

Assignment 3

Assignment 4

Assignment 5

Assignment 6

Assignment 7

Assignment 8

Lecture notes for your help (If you find any typo, please let me know)

Lecture Notes 1 : Introduction

Lecture Notes 2 : Simple Linear Regression Analysis

Lecture Notes 3 : Multiple Linear Regression Model

Lecture Notes 4 : Model Adequacy Checking

Lecture Notes 5 : Transformation and Weighting to Correct Model Inadequacies

Lecture Notes 6  : Diagnostic for Leverage and Influence

Lecture Notes 7  : Generalized and Weighted Least Squares Estimation

Lecture Notes 8  : Indicator Variables

Lecture Notes 9  : Multicollinearity

Lecture Notes 10  : Heteroskedasticity

Lecture Notes 11  :  Autocorrelation

Lecture Notes 12  : Polynomial Regression Models

Lecture Notes 13  : Variable Selection and Model Building

Lecture Notes 14  : Logistic Regression Models

Lecture Notes 15  : Poisson Regression Models

Lecture Notes 16  : Generalized Linear Models

Browse Course Material

Course info.

• Prof. Dimitris Bertsimas

Departments

• Sloan School of Management

As Taught In

• Operations Management
• Probability and Statistics

Learning Resource Types

The analytics edge, 2 linear regression.

2.1 Welcome to Unit 2

• 2.1.1 Welcome to Unit 2

2.2 The Statistical Sommelier: An Introduction to Linear Regression

• 2.2.1 Video 1: Predicting the Quality of Wine
• 2.2.2 Quick Question
• 2.2.3 Video 2: One-Variable Linear Regression
• 2.2.4 Quick Question
• 2.2.5 Video 3: Multiple Linear Regression
• 2.2.6 Quick Question
• 2.2.7 Video 4: Linear Regression in R
• 2.2.8 Quick Question
• 2.2.9 Video 5: Understanding the Model
• 2.2.10 Quick Question
• 2.2.11 Video 6: Correlation and Multicollinearity
• 2.2.12 Quick Question
• 2.2.13 Video 7: Making Predictions
• 2.2.14 Quick Question
• 2.2.15 Video 8: Comparing the Model to the Experts

2.3 Moneyball: The Power of Sports Analytics

• 2.3.1 A Quick Introduction to Baseball
• 2.3.2 Video 1: The Story of Moneyball
• 2.3.3 Video 2: Making it to the Playoffs
• 2.3.4 Quick Question
• 2.3.5 Video 3: Predicting Runs
• 2.3.6 Quick Question
• 2.3.7 Video 4: Using the Models to Make Predictions
• 2.3.8 Quick Question
• 2.3.9 Video 5: Winning the World Series
• 2.3.10 Quick Question
• 2.3.11 Video 6: The Analytics Edge in Sports
• 2.3.12 Quick Question

2.4 Playing Moneyball in the NBA (Recitation)

• 2.4.1 Welcome to Recitation 2
• 2.4.2 Video 1: The Data
• 2.4.3 Video 2: Playoffs and Wins
• 2.4.4 Video 3: Points Scored
• 2.4.5 Video 4: Making Predictions

2.5 Assignment 2

• 2.5.1 Climate Change
• 2.5.2 Reading Test Scores
• 2.5.3 Detecting Flu Epidemics via Search Engine Query Data
• 2.5.4 State Data

Back: 1.5 Assignment Internet Privacy Poll

Welcome to Unit 2

• Download video
• Download transcript

Video 1: Predicting the Quality of Wine

The slides from all videos in this Lecture Sequence can be downloaded here:  Introduction to Linear Regression (PDF - 1.3MB) .

Continue: Quick Question

Introduction to Baseball Video

If you are unfamiliar with the game of baseball, please watch this short video clip for a quick introduction to the game. You don’t need to be a baseball expert to understand this lecture, but basic knowledge of the game will be helpful to you.

TruScribe. “Baseball Rules of Engagement.” March 27, 2012. YouTube. This video is from TrueScribeVideos  and is not covered by our Creative Commons license .

• Back: Video 8: Comparing the Model to the Experts
• Continue: Video 1: The Story of Moneyball

Welcome to Recitation 2

• Back: Quick Question
• Continue: Video 1: The Data

Climate Change

There have been many studies documenting that the average global temperature has been increasing over the last century. The consequences of a continued rise in global temperature will be dire. Rising sea levels and an increased frequency of extreme weather events will affect billions of people.

In this problem, we will attempt to study the relationship between average global temperature and several other factors.

The file climate_change (CSV)  contains climate data from May 1983 to December 2008. The available variables include:

• Year : the observation year.
• Month : the observation month.
• Temp : the difference in degrees Celsius between the average global temperature in that period and a reference value. This data comes from the Climatic Research Unit at the University of East Anglia .
• CO2 ,  N2O , CH4 ,  CFC.11 , CFC.12 : atmospheric concentrations of carbon dioxide (CO2), nitrous oxide (N2O), methane  (CH4), trichlorofluoromethane (CCl3F; commonly referred to as CFC-11) and dichlorodifluoromethane (CCl2F2; commonly referred to as CFC-12), respectively. This data comes from the ESRL/NOAA Global Monitoring Division .
• CO2, N2O and CH4 are expressed in ppmv (parts per million by volume  – i.e., 397 ppmv of CO2 means that CO2 constitutes 397 millionths of the total volume of the atmosphere)
• CFC.11 and CFC.12 are expressed in ppbv (parts per billion by volume). 
• Aerosols : the mean stratospheric aerosol optical depth at 550 nm. This variable is linked to volcanoes, as volcanic eruptions result in new particles being added to the atmosphere, which affect how much of the sun’s energy is reflected back into space. This data is from the Godard Institute for Space Studies at NASA .
• TSI : the total solar irradiance (TSI) in W/m2 (the rate at which the sun’s energy is deposited per unit area). Due to sunspots and other solar phenomena, the amount of energy that is given off by the sun varies substantially with time. This data is from the SOLARIS-HEPPA project website .  
• MEI : multivariate El Nino Southern Oscillation index (MEI), a measure of the strength of the El Nino/La Nina-Southern Oscillation (a weather effect in the Pacific Ocean that affects global temperatures). This data comes from the ESRL/NOAA Physical Sciences Division .

Problem 1.1 - Creating Our First Model

We are interested in how changes in these variables affect future temperatures, as well as how well these variables explain temperature changes so far. To do this, first read the dataset climate_change.csv into R.

Then, split the data into a training set , consisting of all the observations up to and including 2006, and a testing set consisting of the remaining years (hint: use subset). A training set refers to the data that will be used to build the model (this is the data we give to the lm() function), and a testing set refers to the data we will use to test our predictive ability.

Next, build a linear regression model to predict the dependent variable Temp, using MEI, CO2, CH4, N2O, CFC.11, CFC.12, TSI, and Aerosols as independent variables ( Year and Month should NOT be used in the model). Use the training set to build the model.

Enter the model R2 (the “Multiple R-squared” value):

 Numerical Response 

Explanation

First, read in the data and split it using the subset command:

climate = read.csv(“climate_change.csv”)

train = subset(climate, Year <= 2006)

test = subset(climate, Year > 2006)

Then, you can create the model using the command:

climatelm = lm(Temp ~ MEI + CO2 + CH4 + N2O + CFC.11 + CFC.12 + TSI + Aerosols, data=train)

Lastly, look at the model using summary(climatelm). The Multiple R-squared value is 0.7509.

CheckShow Answer

Problem 1.2 - Creating Our First Model

Which variables are significant in the model? We will consider a variable signficant only if the p-value is below 0.05. (Select all that apply.)

If you look at the model we created in the previous problem using summary(climatelm), all of the variables have at least one star except for CH4 and N2O. So MEI, CO2, CFC.11, CFC.12, TSI, and Aerosols are all significant.

Problem 2.1 - Understanding the Model

Current scientific opinion is that nitrous oxide and CFC-11 are greenhouse gases: gases that are able to trap heat from the sun and contribute to the heating of the Earth. However, the regression coefficients of both the N2O and CFC-11 variables are negative , indicating that increasing atmospheric concentrations of either of these two compounds is associated with lower global temperatures.

Which of the following is the simplest correct explanation for this contradiction?

 Climate scientists are wrong that N2O and CFC-11 are greenhouse gases - this regression analysis constitutes part of a disproof. 

 There is not enough data, so the regression coefficients being estimated are not accurate. 

 All of the gas concentration variables reflect human development - N2O and CFC.11 are correlated with other variables in the data set. 

The linear correlation of N2O and CFC.11 with other variables in the data set is quite large. The first explanation does not seem correct, as the warming effect of nitrous oxide and CFC-11 are well documented, and our regression analysis is not enough to disprove it. The second explanation is unlikely, as we have estimated eight coefficients and the intercept from 284 observations.

Problem 2.2 - Understanding the Model

Compute the correlations between all the variables in the training set. Which of the following independent variables is N2O highly correlated with (absolute correlation greater than 0.7)? Select all that apply.

Which of the following independent variables is CFC.11 highly correlated with? Select all that apply.

You can calculate all correlations at once using cor(train) where train is the name of the training data set.

Problem 3 - Simplifying the Model

Given that the correlations are so high, let us focus on the N2O variable and build a model with only MEI, TSI, Aerosols and N2O as independent variables. Remember to use the training set to build the model.

Enter the coefficient of N2O in this reduced model:

(How does this compare to the coefficient in the previous model with all of the variables?)

Enter the model R2:

We can create this simplified model with the command:

LinReg = lm(Temp ~ MEI + N2O + TSI + Aerosols, data=train)

You can get the coefficient for N2O and the model R-squared by typing summary(LinReg).

We have observed that, for this problem, when we remove many variables the sign of N2O flips. The model has not lost a lot of explanatory power (the model R2 is 0.7261 compared to 0.7509 previously) despite removing many variables. As discussed in lecture, this type of behavior is typical when building a model where many of the independent variables are highly correlated with each other. In this particular problem many of the variables (CO2, CH4, N2O, CFC.11 and CFC.12) are highly correlated, since they are all driven by human industrial development.

• Back: Video 4: Making Predictions
• Continue: Reading Test Scores

You are leaving MIT OpenCourseWare

• Mathematics
• Regression Analysis (Video) 
• Co-ordinated by : IIT Kharagpur
• Available from : 2012-07-11
• Intro Video
• Simple Linear Regression
• Simple Linear Regression (Contd.)
• Simple Linear Regression (Contd. )
• Simple Linear Regression ( Contd.)
• Simple Linear Regression ( Contd. )
• Multiple Linear Regression
• Multiple Linear Regression (Contd.)
• Multiple Linear Regression (Contd. )
• Multiple Linear Regression ( Contd.)
• Selecting the BEST Regression Model
• Selecting the BEST Regression Model (Contd.)
• Selecting the BEST Regression Model (Contd. )
• Selecting the BEST Regression Model ( Contd.)
• Multicollinearity
• Multicollinearity (Contd.)
• Multicollinearity ( Contd.)
• Model Adequacy Checking
• Model Adequacy Checking (Contd.)
• Model Adequacy Checking ( Contd.)
• Test for Influential Observations
• Transformation and Weighting to correct model inadequacies
• Transformation and Weighting to correct model inadequacies (Contd.)
• Transformation and Weighting to correct model inadequacies ( Contd.)
• Dummy Variables
• Dummy Variables (Contd.)
• Dummy Variables (Contd. )
• Polynomial Regression Models
• Polynomial Regression Models (Contd.)
• Polynomial Regression Models (Contd. )
• Generalized Linear Models
• Generalized Linear Models (Contd.)
• Non-Linear Estimation
• Regression Models with Autocorrelated Errors
• Regression Models with Autocorrelated Errors (Contd.)
• Measurement Errors and Calibration Problem
• Tutorial - I
• Tutorial - II
• Tutorial - III
• Tutorial - IV
• Tutorial - V
• Watch on YouTube
• Assignments
• Download Videos
• Transcripts
• Self Evaluation (3)