No sé el nombre de esta distribución, pero puedes derivarla de la ley de probabilidad total. Supongamos que tienen distribuciones binomiales negativas con parámetros ( r 1 , p 1 ) y ( r 2 , p 2 ) , respectivamente. Estoy usando la parametrización donde X , Y representan el número de éxitos antes de las fallas r 1 'th y r 2 ' th, respectivamente. Luego,X,Y(r1,p1)(r2,p2)X,Yr1r2
P(X−Y=k)=EY(P(X−Y=k))=EY(P(X=k+Y))=∑y=0∞P(Y=y)P(X=k+y)
Sabemos
P(X=k+y)=(k+y+r1−1k+y)(1−p1)r1pk+y1
y
P(Y=y)=(y+r2−1y)(1−p2)r2py2
entonces
P(X−Y=k)=∑y=0∞(y+r2−1y)(1−p2)r2py2⋅(k+y+r1−1k+y)(1−p1)r1pk+y1
Eso no es bonito (¡ay!). La única simplificación que veo de inmediato es
pk1(1−p1)r1(1−p2)r2∑y=0∞(p1p2)y(y+r2−1y)(k+y+r1−1k+y)
que sigue siendo bastante feo. No estoy seguro de si esto es útil, pero también se puede volver a escribir como
pk1(1−p1)r1(1−p2)r2(r1−1)!(r2−1)!∑y=0∞(p1p2)y(y+r2−1)!(k+y+r1−1)!y!(k+y)!
p-values
I verified with simulation that the above calculation is correct. Here is a crude R function to calculate this mass function and carry out a few simulations
f = function(k,r1,r2,p1,p2,UB)
{
S=0
const = (p1^k) * ((1-p1)^r1) * ((1-p2)^r2)
const = const/( factorial(r1-1) * factorial(r2-1) )
for(y in 0:UB)
{
iy = ((p1*p2)^y) * factorial(y+r2-1)*factorial(k+y+r1-1)
iy = iy/( factorial(y)*factorial(y+k) )
S = S + iy
}
return(S*const)
}
### Sims
r1 = 6; r2 = 4;
p1 = .7; p2 = .53;
X = rnbinom(1e5,r1,p1)
Y = rnbinom(1e5,r2,p2)
mean( (X-Y) == 2 )
[1] 0.08508
f(2,r1,r2,1-p1,1-p2,20)
[1] 0.08509068
mean( (X-Y) == 1 )
[1] 0.11581
f(1,r1,r2,1-p1,1-p2,20)
[1] 0.1162279
mean( (X-Y) == 0 )
[1] 0.13888
f(0,r1,r2,1-p1,1-p2,20)
[1] 0.1363209
I've found the sum converges very quickly for all of the values I tried, so setting UB higher than 10 or so
is not necessary. Note that R's built in rnbinom function parameterizes the negative binomial in terms of
the number of failures before the r'th success, in which case you'd need to replace all of the p1,p2's
in the above formulas with 1−p1,1−p2 for compatibility.