Study of the cost distributions found by Bayes

>> clear
>> load info_posreal_zonapeque
>> prob=infoarchivos2prob(infoarchivos,1:6);

Parameters of the most likely value:

>> [m,ind_max]=max(prob(:));
>> [pot_max,alfa_max,beta_max,lambda_max,mu_max,tipo_max]=ind2sub(size(prob),ind_max);

Now, we will study all values with a relatively high probability, but only those with alfa, beta and pot equal to the most likely ones:

>> ind=find(prob(:)>m/100);
>> [c_pot,c_alfa,c_beta,c_lambda,c_mu,c_tipo]=ind2sub(size(prob),ind);
>> ind=ind(c_pot==pot_max & c_alfa==alfa_max & c_beta==beta_max);
>> [loglambda,logmu,tipo]=infoarchivos2parametrosdistrib(infoarchivos,ind)
>> x=0:.001:4;
>> P=parametrosdistrib2prob(x,loglambda,logmu,tipo);
>> plot(x,P)

>> semilogy(x,P)

>> loglog(x,P)

Very similar, although coming from different types:

>> tipo

 

tipo =

 

1.0000 -1.0000

2.1000 1.0000

2.1000 2.0000

2.1000 2.0000

2.2000 1.0000

2.2000 1.0000

2.2000 1.0000

2.2000 1.0000

2.2000 1.0000

Advertisement