9. Residual Analysis in Linear Regression

Опубликовано: 08 Апрель 2015
на канале: Learn Analytics

11,641

The lecture covers errors/residuals in a linear regression model including analysis. The assumptions of OLS require normally distributed errors with mean of 0 and iid. The lecture covers the SAS syntax in PROC REG to generated the residuals and perform analysis using PROC UNIVARIATE to test normality and basic plots to check for constant variance – homoscedasticity.
Finally the lecture recaps all the steps in doing regression analysis including the various goodness of fit parameters which can be checked for appropriateness of a model.
An understanding of basic statistics including hypothesis testing, variable distributions, P value interpretation is a must to grasp the concepts discusses in this series. Please refer to the Basic Statistics lecture series - http://goo.gl/BqhMqs in order to cover these topics from ground up.
This playlist provides approximately 5 hours of our Analytics Training video series, the training covers Multiple Linear Regression using tools such as SAS, SPSS, Statistica and R using Rattle. For more information, please visit www.learnanalytics.in . For enquiries drop an email to [email protected] .
The objective of the training series is to prepare the student for a career in Data Analysis and the Analytics Industry in general. Please visit our website for further details.
If you wish to subscribe to our full Analytics Training module, please visit http://goo.gl/nIJJHg for our paid Youtube Channel which contains additional hours covering more extensive topics including Linear Regression/Logistic Regression, building and testing predictive models using Logistic / Decision Trees and Ensembling. All videos on our paid channel are available without advertisement interruptions and you can enrol for a 14 day free subscription trial. You pay only if you want to continue.
Additionally, you get access to the datasets discussed in the videos and all SAS Codes.