How to print multiple linear regression on the same scatterplot

[Federico Micol]

Good morning everyone, for a research that I'm performing I'm trying to Plot multiple regression line on the same scatterplot for a Panel dataset, like the one here attached.

Do someone knows how to create these useful representations?
Thanks in advance for the advices.

[George Ford]

something like this work?

twoway scatter tcost output if airline==1 || lfit tcost output if airline==1 || scatter tcost output if airline==2 || lfit tcost output if airline==2

you'll have to gussy it up with colors (mcolor, lcolor, etc..)

[Nick Cox]

Just to flag that the title is misleading here. A multiple regression is one with many predictors. Your problem appears to be showing several regressions with one predictor on the same plot.

[Roman Mostazir]

Extending#2, you are after something like this (fictitious data):

//Fictitious data example:

 li in 1/6, noobs clean

           al         tc         ind  
    Airline 1   14.52905    .1517767  
    Airline 1   14.67648   -.0266089  
    Airline 1   14.38981    .4835709  
    Airline 1   15.14451    .3083085  
    Airline 1   14.12012    .2327023  
    Airline 1   14.28547    .5754296


//Graph:

#delimit ;
twoway (scatter tc ind if al==1, msym(o) mcol(maroon)) ||
       (lfit tc ind if al ==1, lcol(maroon) lpatt(solid) lw(medthick)) ||
       (scatter tc ind if al==2, msym(oh) mcol(green)) ||
       (lfit tc ind if al ==2, lcol(green) lpatt(solid) lw(medthick)) ||
       (scatter tc ind if al==3, msym(t) mcol(yellow)) ||
       (lfit tc ind if al ==3, lcol(yellow) lpatt(solid) lw(medthick)) ||
       (scatter tc ind if al==4, msym(t) mcol(blue)) ||
       (lfit tc ind if al==4, lcol(blue) lpatt(solid) lw(medthick)) ||
       (lfit tc ind, lcol(black) lw(thick)),
       xscale(range(-2.5 1))
      yti("Total Cost", col(black) size(medium))
      legend(off)
       text(10 -.5 "Regression Line", col(black) si(small))
       text(14 0.5 "Airline 1", col(black) si(small))
       text(14 -0.5 "Airline 2", col(black) si(small))
       text(12 -1 "Airline 3", col(black) si(small))
       text(6 -1.5 "Airline 4", col(black) si(small))
       note("{it:Source: Fictitous data}")
;
#delimit cr

[Federico Micol]

Just to flag that the title is misleading here. A multiple regression is one with many predictors. Your problem appears to be showing several regressions with one predictor on the same plot.

Sorry Nick Cox, you are right I should write it better. I was meaning to printi different linear regressions on the same graph. I will change it.

[Federico Micol]

something like this work?

twoway scatter tcost output if airline==1 || lfit tcost output if airline==1 || scatter tcost output if airline==2 || lfit tcost output if airline==2

you'll have to gussy it up with colors (mcolor, lcolor, etc..)

George Ford, thank you for your reply. I will try to set it up. I'm a novice in stata so I'm still experimenting!

[Federico Micol]

Extending#2, you are after something like this (fictitious data):

//Fictitious data example:

li in 1/6, noobs clean

al tc ind
Airline 1 14.52905 .1517767
Airline 1 14.67648 -.0266089
Airline 1 14.38981 .4835709
Airline 1 15.14451 .3083085
Airline 1 14.12012 .2327023
Airline 1 14.28547 .5754296


//Graph:

#delimit ;
twoway (scatter tc ind if al==1, msym(o) mcol(maroon)) ||
(lfit tc ind if al ==1, lcol(maroon) lpatt(solid) lw(medthick)) ||
(scatter tc ind if al==2, msym(oh) mcol(green)) ||
(lfit tc ind if al ==2, lcol(green) lpatt(solid) lw(medthick)) ||
(scatter tc ind if al==3, msym(t) mcol(yellow)) ||
(lfit tc ind if al ==3, lcol(yellow) lpatt(solid) lw(medthick)) ||
(scatter tc ind if al==4, msym(t) mcol(blue)) ||
(lfit tc ind if al==4, lcol(blue) lpatt(solid) lw(medthick)) ||
(lfit tc ind, lcol(black) lw(thick)),
xscale(range(-2.5 1))
yti("Total Cost", col(black) size(medium))
legend(off)
text(10 -.5 "Regression Line", col(black) si(small))
text(14 0.5 "Airline 1", col(black) si(small))
text(14 -0.5 "Airline 2", col(black) si(small))
text(12 -1 "Airline 3", col(black) si(small))
text(6 -1.5 "Airline 4", col(black) si(small))
note("{it:Source: Fictitous data}")
;
#delimit cr

[ATTACH=CONFIG]n1641231[/ATTACH]

Thank you Roman Mostazir, I will try to implement the solutions proposed in this thred.