I make my own script about calculating the transition probability of Markov Chain.I run the two scripts and get different results.Can you detect what makes the difference? What I was doing is [t,j]=size(sets); p00=zeros(j,1); p01=zeros(j,1); p10=zeros(j,1); p11=zeros(j,1); for i=1:j, prev=sets(1:t-1,i); next=sets(2:t,i); p00(i)=sum(prev 1)/sum(sets 1 & next 1); p11(i)=sum(prev > 1 & next > 1)/sum(sets>1); end Another very similiar version is for i=1:nv, s0=sequence(1:nt-1,i); s1=sequence(2:nt,i); ndry=length(find(s0 threshold)); p00(i)=length(find(s0 threshold))/ndry; p10(i)=length(find(s0 > threshold & s1 threshold & s1 > threshold))/nwet; end

Difference in results when running the different version but simple script.

Joy Tian

2015 10 18

0 个回答

3 次查看（30 天）

0 个投票

I make my own script about calculating the transition probability of Markov Chain.I run the two scripts and get different results.Can you detect what makes the difference? What I was doing is

 [t,j]=size(sets);
 p00=zeros(j,1);
 p01=zeros(j,1);
 p10=zeros(j,1);
 p11=zeros(j,1);
 for i=1:j,
    prev=sets(1:t-1,i); 
    next=sets(2:t,i);
    p00(i)=sum(prev <= 1 & next <= 1)/sum(sets<=1);
    p01(i)=sum(prev <= 1 & next > 1)/sum(sets<=1);
    p10(i)=sum(prev > 1 & next <= 1)/sum(sets>1);
    p11(i)=sum(prev > 1 & next > 1)/sum(sets>1);
  end

Another very similiar version is

for i=1:nv,
  s0=sequence(1:nt-1,i);
  s1=sequence(2:nt,i);
  ndry=length(find(s0 <= threshold));
  nwet=length(find(s0 > threshold));
  p00(i)=length(find(s0 <= threshold & s1 <= threshold))/ndry;
  p01(i)=length(find(s0 <= threshold & s1 > threshold))/ndry;
  p10(i)=length(find(s0 > threshold & s1 <= threshold))/nwet;
  p11(i)=length(find(s0 > threshold & s1 > threshold))/nwet;
end