Presentation is loading. Please wait.

Presentation is loading. Please wait.

Multiple Linear Regression Model

Similar presentations


Presentation on theme: "Multiple Linear Regression Model"— Presentation transcript:

1 Multiple Linear Regression Model
Until now, all we did was establish the relationship between 2 variables 1 independent variable, 1 dependent variable But we would like to do better and several independent variables together could be a better predictor, so we would like to establish a relationship like this - Several independent variables, 1 dependent variable

2 Analysis of Relationship among several variables
Linear Regression estimates the “Line of Best Fit” (‘Line’ because we are talking about a Linear model)

3 Analysis of Relationship among several variables
bj for each X is slope coefficient The slope coefficient measures how much the dependent variable, Y , changes when the independent variable, Xj , changes by one unit, holding all other independent variables constant. a is the intercept or constant for the model The intercept measures the value of Y if all X’s are 0 The way to estimate this is through the process of ‘Least Squares’ In practice, software programs are used to estimate the multiple regression model

4 Inference about Parameters
Parameter Estimate Error t Value Pr>|t| CONST X X X X X X X X X X

5 Assumptions of the Multiple Regression Model
The relationship between the dependent variable, Y, and the independent variables, X1, X2, , Xk, is linear. The independent variables (X1, X2, , Xk) are not random. Also, no exact linear relation exists between two or more of the independent variables. The error term is normally distributed The expected value (mean) of the error term, conditioned on the independent variables, is 0 The variance of the error term is the same for all observations

6 Inference about Model If the model is correctly specified, R2 is an ideal measure Addition of a variable to a regression will increase the R2 (by construction) This fact can be exploited to get regressions with R2 ~ 100% by addition of variables, but this doesn’t mean that the model is any good Adj-R2 should be reported

7 Inference about Model Adjusted R2 is a measure of goodness of fit that accounts for additional explanatory variables.

8 Inference about Parameters
Coefficients (b1, b2,..,bk) are estimated with a confidence interval To know if a specific independent variable (xi) is influential in predicting the dependent variable (y), we test whether the corresponding coefficient (bi) is statistically different from 0 (i.e. bi = 0). We do so by calculating the t-statistic for the coefficient If the t-stat is sufficient large, it indicates that bi is significantly different from 0 indicating that bi * xi plays a role in determining y

9 Inference about Parameters
Parameter Estimate Error t Value Pr>|t| CONST X X X X X X X X X X

10 Predicting the Dependent Variable
To predict the value of a dependent variable using a multiple linear regression model, we follow these three steps: Obtain estimates of the regression parameters. Determine the assumed values of the independent variables.

11 Predictions Estimate a model Assess the model predictive ability
Assess the significance of each independent variable If they are satisfactory, we decide to use the existing model Else we re-estimate using different independent variables

12 Predictions We then change the parameters to levels that we would like predictions for derive the corresponding y This would be the predicted value of ‘y’ as a function of the changes to the existing parameter values

13 Using Dummy Variables Some times our independent variable isn’t numeric E.g. establish relationship between day of week and alcohol consumption relationship between major and income More relevant example from finance Industry effect on returns (do technology stocks higher returns on same investments) Do emerging markets provide greater returns for same risks?

14 Using Dummy Variables A dummy variable is qualitative variable
It takes on a value of 1 if a particular condition is true and 0 if that condition is false. In our examples, we would use dummy for the following Relationship between week day and alcohol consumption Each weekday would have a dummy associated if the alcohol consumption on the y corresponded to it

15 Using Dummy Variables

16 Month-of-the-Year Effects on Small Stock Returns
Suppose we want to test whether total returns to one small-stock index, the Russell 2000 Index, differ by month. We can use dummy variables in estimate the following regression,

17 Month-of-the-Year Effects on Small Stock Returns

18 Violations of Regression Assumptions
Inference based on an estimated regression model rests on certain assumptions being met. Violations may cause the inferences made to be invalid.

19 Heteroskedasticity Heteroskedasticity occurs when the variance of the errors differs across observations. does not affect consistency t-tests for the significance of individual regression coefficients are unreliable because heteroskedasticity introduces bias into estimators of the standard error of regression coefficients.

20 Regressions with Homoskedasticity

21 Regressions with Heteroskedasticity

22 Serial Correlation When regression errors are correlated across observations, we say that they are serially correlated (or autocorrelated). Serial correlation most typically arises in time-series regressions. The principal problem caused by serial correlation in a linear regression is that it leads to incorrect calculation of critical values for the test of significance

23 Multicollinearity Multicollinearity occurs when two or more independent variables (or combinations of independent variables) are highly (but not perfectly) correlated with each other. does not affect the consistency of the regression coefficients estimates become extremely imprecise and unreliable The classic symptom of multicollinearity is a high R2 even though the t-statistics on the estimated slope coefficients are not significant. The most direct solution to multicollinearity is excluding one or more of the regression variables.

24 Problems in Regression & Solutions
Effect Solution Heteroskedasticity Incorrect standard errors Correct for conditional heteroskedasticity Serial Correlation Incorrect t-values Correct for serial correlation Multicollinearity High R2 and low t-statistic Remove 1 or more independent variable

25 Model Specification Model specification refers to the set of variables included in the regression and the regression equation’s functional form. Possible misspecifications include: One or more important variables could be omitted from regression. One or more of the regression variables may need to be transformed (for example, by taking the natural logarithm of the variable) before estimating the regression. The regression model pools data from different samples that should not be pooled.

26 Discreet Dependent Variable Models
Discreet dependent variables are dummy variables used as dependent variables instead of as independent variables. Mainly 2 models – The probit model, which is based on the normal distribution, estimates the probability that Y = 1 (a condition is fulfilled) given the value of the independent variable X The logit model is identical, except that it is based on the logistic distribution rather than the normal distribution

27 Discreet Dependent Variable Models
Logistic regression


Download ppt "Multiple Linear Regression Model"

Similar presentations


Ads by Google