What is the history of regression analysis?

ygao

New Member
#1
Hi all,

Recently, I am using regression analysis to do a geophysics study. Now the study is almost done. I start to write my report. I want to add some history of the regression technique itself. But I am far from an expert of statistics. So I have no idea about what I should write.

Therefore, I want to ask whether anyone can recommend me some articles about the history of regression analysis. How it is invented... How it is developed in history... What is the current topic about this technique... etc...

Or if you have suggestions on how I should finish the regression part of my report, it is highly appreciated.

Thx in advance,
yg
 
#2
I wouldn't write about the history of regression... That would seem (to me) irrelevant in almost any work you could imagine (except papers on pure statistics or econometrics). Instead I would recommend you, if you absolutely want to write something about the technique, that you derive the model, meaning that you solve the problem: minimize sum of squared errors. Then calculate derivates with respect to alpha and beta, and solve the model. Maybe do the same, but with matrices, which is much more handy for multiple linear regression analysis. But history is irrelevant I think...
 

TheEcologist

Global Moderator
#3
Hi all,

Recently, I am using regression analysis to do a geophysics study. Now the study is almost done. I start to write my report. I want to add some history of the regression technique itself. But I am far from an expert of statistics. So I have no idea about what I should write.

Therefore, I want to ask whether anyone can recommend me some articles about the history of regression analysis. How it is invented... How it is developed in history... What is the current topic about this technique... etc...

Or if you have suggestions on how I should finish the regression part of my report, it is highly appreciated.

Thx in advance,
yg
The method (of least squares) was first described by Carl Friedrich Gauss around 1794.

According to Carter (Rice University, 1995 - textbook on Linear Algebra), Gauss developed least squares to solve a chemistry problem for his friend (when Gauss was about the age of a high-school senior). Gauss however did not publish the method until 1809.

According to wikipedia:

"The idea of least-squares analysis was also independently formulated by the Frenchman Adrien-Marie Legendre in 1805 and the American Robert Adrain in 1808"

The term regression comes from Francis Galton (though then only used in biology). His student Karl Pearson is then accredited to have moved it to a more general statistical context.

Generally, all the references to the mentioned seminal papers should be available through the wikipedia page.
 
#4
I agree with Riverdale that writing aboutthe history of regression techniques is irrelevant. It's by no means a new idea and it would seem a little strange to explain something that's just standard now. I think deriving the model is also unnecessary unless this is a field where regression is almost never used and your readers might need a reminder of how it works (I really doubt this). If you need to fill out the paper I would take the space you were going to use for the history and use it to elaborate on the robustness of your results--what implicit assumptions have you made in using your particular technique and why? What about your results might change if those assumptions are violated? Are there remaining unanswered questions that further research could get at? Those sorts of things are much more relevant in an academic paper than an out of place history lesson.

Good luck with the rest of your project.
 

TheEcologist

Global Moderator
#5
Those sorts of things are much more relevant in an academic paper than an out of place history lesson
On the other hand, such initiatives are encouraged in students reports. I certainly wouldn't rebuke an undergrad student for showing an interest. Don't jump to conclusions guys.

ygao, if you are writing an academic paper or professional report well then Riverdale27 and AtlasFrysmith are completely spot on. Stay relavent and on topic. If you suspect that regression analysis is completely unknown for your focus group, give only a short elaboration.

However if your not writing an academic paper or professional report , go ahead and do what you think is interesting. As a supervisor, I would not frown on such an initiative (on the contrary).