In statistics, linear regression is a linear approach to modelling the relationship between a scalar response and one or more explanatory variables (also known as dependent and independent variables).
From the data analysis sheet, I found the two figures. Which one is selected for your manuscript and why select 1/2?
If I set the Regression equation, R square, what will you do? Dr Rabin Thapa
Both of them are needed. The first is important to show the lack of correlation between the residuals and the second illustrates the normality of the residuals (without considering the correlation between residuals or the effect of the intended variables on residuals).
Thank you very much for your nicely presented description and wishes you best of luck Dr Babak Jamshidi. Really, thanks Babak Jamshidi
Thank you very much for your nicely presented description and wishes you best of luck Dr Babak Jamshidi. Really, thanks Babak Jamshidi
Both are important but if you use regression equation with R square and one graph is perfect where is set the relation.
I also think so but need prove. I think, both figures are important and both are describe in research paper Dr Agriculturist M. Hassain
Dear author, thank you for sharing this excellent question.
I agree with Babak Jamshidi.
With kind regards.
I guess, both of them are equally important and can be applied to different occasions because of their own features. Their uses may vary depending upon the variables conisdered and set of data available. But in case of first one(Scatter diagram representation), even it shows a relationship, we cannot assume that one variable caused the other instead both may be influenced by a third variable.
2 is usefull to check if the residuals are randomly spread (Gauss distributed), which would confirm your model. If you see a systematic pattern in the residuals, it indicates that another independent variable is influencing the dependent variable.
The two figures tell different stories about the data, so, both are relevant
I think both figures are important and useful as it depends upon how would you like to express your data or results
Ahasan -
While both can be helpful, I'd be concerned about the first figure. Besides heteroscedasticity, which is a feature to be expected, not a bug, my guess is that you might also have a missing variable. There seem to be two groups of estimated residuals. Also, perhaps you missed a polynomial or other such relationship. Perhaps you should revisit your model selection process, and also try some kind of cross-validation. That is, you can see if the model you select based on one particular sample will 'predict' well for another sample.
While you don't want to oversimplify your model, you don't want it overcomplicated either. [And please note that if you get the model close to ideal, I think you are more likely to have heteroscedasticity. (I'm writing a paper on that now.)]
Cheers - Jim
Thank you very much and wishes you very best of luck Dr Muhammad Ihsan
Thank you very much and wishes you very best of luck Dr Dlshad Ali Omer
Thank you very much and wishes you very best of luck Dr Humberto Costa
Thank you very much and wishes you very best of luck Dr Aparna Sathya Murthy
Thank you very much and wishes you very best of luck Dr Krishna Raj Pandey
Thank you very much and wishes you very best of luck Dr ميسم شهاب
Thank you very much and wishes you very best of luck Dr Кhrystova Tеtіana Кhristovaya T.E. Христова
Thank you very much and wishes you very best of luck Dr Koen Van de Moortel
Thank you very much and wishes you very best of luck Dr Maurice Ekpenyong
Thank you very much and wishes you very best of luck Dr Lamia Mustafa Al-Naama
Thank you very much and wishes you very best of luck Dr James R Knaub
- Badges
- Science topic
- Similar topics
- Linguistics
- Philology
- Classical Philology
- Manuscripts