# Multiplying two matrices of probabilities with different sizes.

8 views (last 30 days)
Haya A on 23 Aug 2018
Commented: Haya A on 24 Aug 2018
Hi guys,
I have two matrices, with different sizes,that represents the probabilities of occurrence for two independent events. I want to create a new matrix that represent the probability of occurrence for these two events together. I tried multiplying the matrices,but the sum of the probabilities did not produce 1.
any Ideas? this is the code :
attributesize=4;
FileNumber=10;
BSTotal=7;
Files=exprnd(1,attributesize,FileNumber);
Files=Files./repmat(sum(Files,1),attributesize,1);
BS=exprnd(1,BSTotal,attributesize);
BS=BS./repmat(sum(BS,2),1,attributesize);
R= BS*Files;
Torsten on 23 Aug 2018
Edited: Torsten on 23 Aug 2018
Your matrices contain random numbers from an exponential distribution. What makes you think they contain probabilities of occurence of a certain event ?
And why probabilitie s ? The occurence of an event has one unique probability.
Haya A on 23 Aug 2018
I'm generating random probabilities of occurrence for each attributes, or feature, and the sum of all probabilities are 1. I want to calculate the probability that certain file x with the probability , for example 0.23, for feature m, will be requested by user y with certain probability to the same feature. I want to do it for all.

Kaushik Lakshminarasimhan on 24 Aug 2018
Edited: Kaushik Lakshminarasimhan on 24 Aug 2018
I don't think there is anything wrong with your code. What is wrong is your interpretation of the outcome of multiplying your matrices.
Each entry of R denotes the probability that the file actually contains the attribute that the user requests. Specifically, R(i,j) denotes the probability that file number j contains the attribute that user number i requested. This number need not be 1. Also, the sum of columns or sum of rows of R need not be 1 --- the sums are not even probability values because they represent a sum of probabilities of events that are not mutually exclusive.
Haya A on 24 Aug 2018
Yes, I understand that. But I am trying to generate random requests for these files with R probabilities using randsrc function, and I can't do that if the sum of columns or rows is not 1. do you have suggestion?

### Categories

Find more on t Location-Scale Distribution in Help Center and File Exchange

### Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!