suicide=c(1,25,40,83,123,256,1,27,49,84,126,257,1,27,49,84,129,311,5,30,54,84,
134,314,7,30,56,90,144,322,8,31,56,91,147,369,8,31,62,92,153,415,13,32,63,93,
163,573,14,34,65,93,167,609,14,35,65,103,175,640,17,36,67,103,228,737,18,37,
75,111,231,21,38,76,112,235,21,39,79,119,242,22,39,82,122,256)

hist(suicide,freq=F,breaks=20)
x=seq(min(suicide),max(suicide),l=200)
#exponential distribution fit:
lines(x,dexp(x,rate=1/mean(suicide)),lty=2,col=2)

#UMD fits with estimated m using CN, AIC & BIC:
fit.cn=umd(suicide,fix.lower=0)
fit.cn$m.hat
lines(x,fit.cn$dumd(x))
fit.aic=umd(suicide,,fix.lower=0,crit='AIC',warning=F)
fit.aic$m.hat
lines(x,fit.aic$dumd(x),lty=2)
fit.bic=umd(suicide,,fix.lower=0,crit='BIC',warning=F)
fit.bic$m.hat
lines(x,fit.bic$dumd(x),lty=3)

#UMD fit with maximum m:
m.max=ceiling(length(suicide)/log(length(suicide)))
fit.max=umd(suicide,,fix.lower=0,m=m.max,warning=F)
fit.max$m.hat
lines(x,fit.max$dumd(x),col='blue')

#KS test for different fits:
ks.test(suicide,pexp,rate=1/mean(suicide))
ks.test(suicide,fit.cn$pumd)
ks.test(suicide,fit.aic$pumd)
ks.test(suicide,fit.bic$pumd)
ks.test(suicide,fit.max$pumd)