Ordinary Least Squares regression or Linear regression

welcome to this excel stat short video on linear regression in this presentation I will explain to you how ordinary least-squares regression functions this type of regression is also sometimes referred to as linear regression if we call X the explanatory variable and Y the dependent variable the linear model writes that the Y value for a sample in is the sum of its values on X multiplied by the beta coefficients plus an error the way the beta coefficients are found is by minimizing the errors hence the name ordinary least-squares regression in the case of the simple linear regression where we have only one X variable it is very easy to visualize the regression line and the errors of prediction which are the difference between their real value and the predicted value in the case of multiple linear regression it gets a bit more complex but we will use a visual example to illustrate it we can take two X variables and as you see we can find a regression plane that minimizes the error of prediction however we should stress that ordinary least-squares regression has some limitations firstly the redundant information in the explanatory variables that we can measure by their glynnie arity can lead to a misinterpretation of the model and the beta coefficients then it is necessary to have more observations than the number of X variables finally only one response variable can be modeled to overcome the first two issues : e era t between the X variables and the necessity of more samples and variables we can use principal component regression partial least squares regression also solves these issues but in addition enables you to have several dependent variables

Channel: Addinsoft

Views: 142,633

Rating: 4.5040431 out of 5

Keywords: linear regression, multiple linear regression, linear regression software

Id: VZkXjh0m1Rg

Length: 2min 10sec (130 seconds)

Published: Tue Mar 29 2011