Can Stata draw curved line, even manually ?

[Lisa Wilson]

Dear all
Is there a way to draw a curved line in a graph in Stata, even manually ?
I want to draw a business cycle similar to the one in the attached graph. My aim is to draw the typical business cycle in a graph and then incorporate my actual data (that i have found to mimic the business cycle turning points)...The problem now is with drawing this business cycle.

I tried to draw a grpah in stata and then draw a curved line similar to the business cycle after opening the graph editor. Unfortunately, I found only straight lines available to be incorporated into the graph.

Thanks for your help
Lisa

[Scott Merryman]

Like this?

clear
set obs 1000
gen x = _n
gen y = 100*sin(x/100) + .25*x + 100
line y x

[Dimitriy V. Masterov]

Another option would be to smooth the data or even fit a curve to it:

webuse sales1, clear
tssmooth ma sm1=sales, window(2 1 2 )
tw (connected sales t) (line sm1 t) (lpoly sales t)

[Leonardo Guizzetti]

Here is a cool way to graph any separable function, y=f(x). The example below approximates the curve specified in your drawing, but you can edit what comes after function to be anything. It also uses twoway as suggested by Dimitriy.

graph twoway (function y=sin(3*x*3.14159)+x)

[Lisa Wilson]

Dear all
Thanks a lot for your suggestions.I find the one in #4 by Leonardo helpful and generic.

The problem now is that I want this curve to be incorporated in the graph I have attached in this current post, where the peak in the curve (in #1) parallels the event year in my new graph (in post #5). Also, another graph where the trough parallels the event year.

Any help is much appreciated.
Thanks you so much.
Lisa

[Lisa Wilson]

Dear all
To clarify I use this code to produce the graph in #5

gen peak1=0
replace peak1=1 if year==1973&qtr==4

gen peak2=0
replace peak2=1 if year==1980&qtr==1

gen peak3=0
replace peak3=1 if year==1981&qtr==3

gen peak4=0
replace peak4=1 if year==1990&qtr==3

gen peak5=0
replace peak5=1 if year==2001&qtr==1

gen peak6=0
replace peak6=1 if year==2007&qtr==4


egen eventyr1=max(cond(peak1==1, fqdate, .))
gen timesinceevent1= fqdate-eventyr1

egen eventyr2=max(cond(peak2==1, fqdate, .))
gen timesinceevent2= fqdate-eventyr2

egen eventyr3=max(cond(peak3==1, fqdate, .))
gen timesinceevent3= fqdate-eventyr3

egen eventyr4=max(cond(peak4==1, fqdate, .))
gen timesinceevent4= fqdate-eventyr4

egen eventyr5=max(cond(peak5==1, fqdate, .))
gen timesinceevent5= fqdate-eventyr5

egen eventyr6=max(cond(peak6==1, fqdate, .))
gen timesinceevent6= fqdate-eventyr6

twoway (scatter shock timesinceevent1 if inrange(timesinceevent1, -4, 4), connect(l) sort) ///
(scatter shock timesinceevent2 if inrange(timesinceevent2, -4, 4), connect(l) sort) ///
(scatter shock timesinceevent3 if inrange(timesinceevent3, -4, 4), connect(l) sort) ///
(scatter shock timesinceevent4 if inrange(timesinceevent4, -4, 4), connect(l) sort) ///
(scatter shock timesinceevent5 if inrange(timesinceevent5, -4, 4), connect(l) sort) ///
(scatter shock timesinceevent6 if inrange(timesinceevent6, -4, 4), connect(l) sort)

Does anyone know how to draw incorporate a curve similar to #1 where the peak parallels the event year?

Thank you

[Lisa Wilson]

Dear all
This is quite old now.
Any suggestion here?

The code used in #6 shows the behavior of the variable "shock" four quarters before and after peak1, peak2...peak6.
All what I need is to incorporate the actual cycle as in #1 ??

[ericmelse]

For those visitors interested in the above, do visit the stata-graphfunctions Github page of Asjad Naqvi that offers a suite of programs to support Stata visualization customizations.
The program catspline allows users to generate splines based on the Catmull-Rom algorithm.