Regression Analysis if anyone would like to give me a personal brief explanation (Including the Statistics portion of it)

Question

anonymous · Answer

You know that this is a huge chapter right?

katrinakaif · Answer

I do, Just need a very quick explanation for a review.

anonymous · Answer

Lol, Good luck :)

mani_jha · Answer

The word 'regression' means 'going back to/returning'. Suppose you're assigned to 'predict' the total number of applicants that will be selected in MIT from 5000 applications in 2012. 
For such a task, you would want to know what has been the selection rate in the past few years. You're 'regressing' or 'going back' to the past records so that you can make a suitable prediction for this year.
You find that in the previous years 1000 out of 6000(2009), 2000 out of 8000(2010), 2500 out of 9000(2011). There must be some relation between these two quantities - the total no of applications, and the number of applications selected. You plot a graph between these two quantities.
You may get something like this:, if you plot for several years:
|dw:1332610548396:dw|
This is a scattered plot. You can barely predict anything out of this. That's why you draw a straight line closest to these points, whose equation you can write and then easily predict the no of selections by putting x=5000. This line is known as the best fitting line, or the line of regression..
y=a+bx
b is known as the regression coefficient.
There  are several formulae for the best fitting line, regression coefficient etc but for that you need to known correlations. Do you know that?
P.S. I think it is best for you to not start anything new just before the examination. Try to do your best in what you already know.

mani_jha · Answer

|dw:1332610950098:dw|
See that the line of regression is the closest possible line to your points.