Dot plot help

[Jack Bowley]

Hello, Can anyone advise on how to code a dotplot to produce similar results to the graph shown below. I have panel data with similar variables to that detailed in the graph but cannot understand how to replicate using my own data in stata using statas command for dotplots.

[Andrew Musau]

help graph dot

We cannot really do any better without a data example. See FAQ Advice #12 for details.

[Jack Bowley]

help graph dot

We cannot really do any better without a data example. See FAQ Advice #12 for details.

Sorry for being too vague. I have a panel dataset in which I'm hoping to calculate the effect of earthquakes on political score in developing countries. I would like to insert a dot plot similar to that as shown above that can help visualise the # of earthquakes per year across each Country (or a smaller sample as i have 118 countries). The size of the dot indicates the number of earthquakes a country experienced in a given year. More earthquakes = Bigger dot.

I have tried following the dot plot help guide, but I'm just guessing and don't really understand it

graph dot NoOfQuakes ascategory(Year), over(Cntry)

I understand this is probably nonsense and is a long way off what I'm after.

Essentially, I would like a dot matrix with my countries on the Y axis, years on the X axis, and each dot representing an earthquake in a given country in a given year, likewise More earthquakes=Bigger dot

A sample of my data is:

* Example generated by -dataex-. To install: ssc install dataex
clear
input str24 Country int Year byte NoOfQuakes
"Albania" 1960 .
"Albania" 1961 .
"Albania" 1962 .
"Albania" 1963 .
"Albania" 1964 .
"Albania" 1965 .
"Albania" 1966 .
"Albania" 1967 .
"Albania" 1968 .
"Albania" 1969 .
"Albania" 1970 .
"Albania" 1971 .
"Albania" 1972 .
"Albania" 1973 .
"Albania" 1974 .
"Albania" 1975 .
"Albania" 1976 .
"Albania" 1977 .
"Albania" 1978 .
"Albania" 1979 .
"Albania" 1980 .
"Albania" 1981 .
"Albania" 1982 .
"Albania" 1983 .
"Albania" 1984 .
"Albania" 1985 .
"Albania" 1986 .
"Albania" 1987 .
"Albania" 1988 .
"Albania" 1989 .
"Albania" 1990 .
"Albania" 1991 .
"Albania" 1992 .
"Albania" 1993 .
"Albania" 1994 .
"Albania" 1995 .
"Albania" 1996 .
"Albania" 1997 .
"Albania" 1998 .
"Albania" 1999 .
"Albania" 2000 .
"Albania" 2001 .
"Albania" 2002 .
"Albania" 2003 .
"Albania" 2004 .
"Albania" 2005 .
"Albania" 2006 .
"Albania" 2007 .
"Albania" 2008 .
"Albania" 2009 .
"Albania" 2010 .
"Albania" 2011 .
"Albania" 2012 .
"Albania" 2013 .
"Albania" 2014 .
"Albania" 2015 .
"Albania" 2016 .
"Albania" 2017 .
"Albania" 2018 .
"Algeria" 1960 1
"Algeria" 1961 .
"Algeria" 1962 .
"Algeria" 1963 .
"Algeria" 1964 .
"Algeria" 1965 1
"Algeria" 1966 .
"Algeria" 1967 1
"Algeria" 1968 .
"Algeria" 1969 .
"Algeria" 1970 .
"Algeria" 1971 .
"Algeria" 1972 .
"Algeria" 1973 .
"Algeria" 1974 .
"Algeria" 1975 .
"Algeria" 1976 .
"Algeria" 1977 .
"Algeria" 1978 .
"Algeria" 1979 .
"Algeria" 1980 1
"Algeria" 1981 .
"Algeria" 1982 .
"Algeria" 1983 .
"Algeria" 1984 .
"Algeria" 1985 1
"Algeria" 1986 .
"Algeria" 1987 1
"Algeria" 1988 1
"Algeria" 1989 1
"Algeria" 1990 .
"Algeria" 1991 .
"Algeria" 1992 1
"Algeria" 1993 .
"Algeria" 1994 1
"Algeria" 1995 .
"Algeria" 1996 .
"Algeria" 1997 .
"Algeria" 1998 .
"Algeria" 1999 1
"Algeria" 2000 .
"Algeria" 2001 .
"Algeria" 2002 .
"Algeria" 2003 2
"Algeria" 2004 2
"Algeria" 2005 .
"Algeria" 2006 1
"Algeria" 2007 .
"Algeria" 2008 .
"Algeria" 2009 .
"Algeria" 2010 .
"Algeria" 2011 .
"Algeria" 2012 .
"Algeria" 2013 .
"Algeria" 2014 1
"Algeria" 2015 .
"Algeria" 2016 .
"Algeria" 2017 .
"Algeria" 2018 .
"Angola"  1960 .
"Angola"  1961 .
"Angola"  1962 .
"Angola"  1963 .
"Angola"  1964 .
"Angola"  1965 .
"Angola"  1966 .
end

[Andrew Musau]

You can use twoway scatter. Probably you would want to sort the axis in terms of frequency of earthquakes. Below, I use labmask from the Stata Journal to do this. Change the highlighted numbers to correspond to the number of countries. In this case, the data example has only three.

* Example generated by -dataex-. To install: ssc install dataex
clear
input str24 Country int Year byte NoOfQuakes
"Albania" 1960 .
"Albania" 1961 .
"Albania" 1962 .
"Albania" 1963 .
"Albania" 1964 .
"Albania" 1965 .
"Albania" 1966 .
"Albania" 1967 .
"Albania" 1968 .
"Albania" 1969 .
"Albania" 1970 .
"Albania" 1971 .
"Albania" 1972 .
"Albania" 1973 .
"Albania" 1974 .
"Albania" 1975 .
"Albania" 1976 .
"Albania" 1977 .
"Albania" 1978 .
"Albania" 1979 .
"Albania" 1980 .
"Albania" 1981 .
"Albania" 1982 .
"Albania" 1983 .
"Albania" 1984 .
"Albania" 1985 .
"Albania" 1986 .
"Albania" 1987 .
"Albania" 1988 .
"Albania" 1989 .
"Albania" 1990 .
"Albania" 1991 .
"Albania" 1992 .
"Albania" 1993 .
"Albania" 1994 .
"Albania" 1995 .
"Albania" 1996 .
"Albania" 1997 .
"Albania" 1998 .
"Albania" 1999 .
"Albania" 2000 .
"Albania" 2001 .
"Albania" 2002 .
"Albania" 2003 .
"Albania" 2004 .
"Albania" 2005 .
"Albania" 2006 .
"Albania" 2007 .
"Albania" 2008 .
"Albania" 2009 .
"Albania" 2010 .
"Albania" 2011 .
"Albania" 2012 .
"Albania" 2013 .
"Albania" 2014 .
"Albania" 2015 .
"Albania" 2016 .
"Albania" 2017 .
"Albania" 2018 .
"Algeria" 1960 1
"Algeria" 1961 .
"Algeria" 1962 .
"Algeria" 1963 .
"Algeria" 1964 .
"Algeria" 1965 1
"Algeria" 1966 .
"Algeria" 1967 1
"Algeria" 1968 .
"Algeria" 1969 .
"Algeria" 1970 .
"Algeria" 1971 .
"Algeria" 1972 .
"Algeria" 1973 .
"Algeria" 1974 .
"Algeria" 1975 .
"Algeria" 1976 .
"Algeria" 1977 .
"Algeria" 1978 .
"Algeria" 1979 .
"Algeria" 1980 1
"Algeria" 1981 .
"Algeria" 1982 .
"Algeria" 1983 .
"Algeria" 1984 .
"Algeria" 1985 1
"Algeria" 1986 .
"Algeria" 1987 1
"Algeria" 1988 1
"Algeria" 1989 1
"Algeria" 1990 .
"Algeria" 1991 .
"Algeria" 1992 1
"Algeria" 1993 .
"Algeria" 1994 1
"Algeria" 1995 .
"Algeria" 1996 .
"Algeria" 1997 .
"Algeria" 1998 .
"Algeria" 1999 1
"Algeria" 2000 .
"Algeria" 2001 .
"Algeria" 2002 .
"Algeria" 2003 2
"Algeria" 2004 2
"Algeria" 2005 .
"Algeria" 2006 1
"Algeria" 2007 .
"Algeria" 2008 .
"Algeria" 2009 .
"Algeria" 2010 .
"Algeria" 2011 .
"Algeria" 2012 .
"Algeria" 2013 .
"Algeria" 2014 1
"Algeria" 2015 .
"Algeria" 2016 .
"Algeria" 2017 .
"Algeria" 2018 .
"Angola"  1960 .
"Angola"  1961 .
"Angola"  1962 .
"Angola"  1963 .
"Angola"  1964 .
"Angola"  1965 .
"Angola"  1966 .
end

bys Country: egen freq= total(NoOfQuakes)
sort freq Country
gen order= sum(Country!= Country[_n-1]) 
labmask order, values(Country)
set scheme s1mono
tw scatter order Year if NoOfQuakes [fw= NoOfQuakes ] , ///
msy(oh) ylab(1/3, val angle(hor))  xlab(1960(5)2015) ytitle("") ///
xtitle("") yline(1/3, lcolor(black) lp(dot)) ysc(r(0.5 3.5))

[Andrew Musau]

Also note that the sample should include countries that have had at least one earthquake. So start with

bys Country (NoOfQuakes): drop if missing(NoOfQuakes[1])

[Jack Bowley]

Also note that the sample should include countries that have had at least one earthquake. So start with

bys Country (NoOfQuakes): drop if missing(NoOfQuakes[1])

Sorry for my delayed reply, I have been occupied with some other things. With a little bit of fiddling this is exactly what i was after. Thankyou very much for your haste response and expertise.

By any chance, would it be possible to take a random selection of countries to use in the graph, as opposed to using all 58 countries that I have data for?

[Andrew Musau]

By any chance, would it be possible to take a random selection of countries to use in the graph, as opposed to using all 58 countries that I have data for?

frame put Country, into(select)
frame select{
    contract Country, nomiss
    gen rand= rnormal()
    sort rand
    *SELECT 20 (CHANGE TO SPECIFIC NUMBER)
    keep in 1/20
}
frlink m:1 Country, frame(select)
keep if !missing(select)
frame drop select

[Nick Cox]

Here is another way to do it without frames. The premise is that we select either all observations or none for each country. Choose your own value of 20.

select is 1 for selected countries.

egen tag = tag(Country)
set seed 314159265
gen rndm = runiform() if tag
sort rndm
gen select = _n <= 20
bysort Country (select) : replace select = select[_N]

[Jack Bowley]

Here is another way to do it without frames. The premise is that we select either all observations or none for each country. Choose your own value of 20.

select is 1 for selected countries.

egen tag = tag(Country)
set seed 314159265
gen rndm = runiform() if tag
sort rndm
gen select = _n <= 20
bysort Country (select) : replace select = select[_N]

Thankyou to you and Andrew for your advice. Sorry for the silly questions, how would I incorporate this code into Andrews aforementioned code so that the final graph uses the random selection code that you have described.

[Nick Cox]

"aforementioned" here I think means #2. If so then the code should start

 
 tw scatter order Year if NoOfQuakes & select

[Jack Bowley]

"aforementioned" here I think means #2. If so then the code should start

tw scatter order Year if NoOfQuakes & select

Hi Nick, can you address where I'm going wrong. So before I use the random code, I use

tw scatter order Year if NoOfQuakes [fw= NoOfQuakes ] ,msy(oh) ylab(1/58, val angle(hor))  xlab(1960(5)2015) ytitle("Country") xtitle("Years") yline(1/58, lcolor(black) lp(dot)) ysc(r(0.5 3.5))

and obtain...

Now after I run your code,

egen tag = tag(Country)

. set seed 314159265

. gen rndm = runiform() if tag
(3,364 missing values generated)

. sort rndm

. gen select = _n <= 20

. bysort Country (select) : replace select = select[_N]
(1,160 real changes made)

. tw scatter order Year if NoOfQuakes & select [fw= NoOfQuakes ] ,msy(oh) ylab(1/58, val angle(hor))  xlab(1960(5)2015) ytitle("Country") xtitle("Years") yline(1/58, lcolor(black) lp(dot)) ysc(r(0.5 3.5))

This isn't changing the graph above. Could you please identify my mistakes.

Thankyou again both Nick and Andrew.

[Nick Cox]

Possibly somethng to do with order and Country.