Making random to work like randi

I have this code

idx = randi(N_clusters,N_clusters,1);

but I want to use the random function to achieve the same output. I do not get into why I want to use the random function but it is important for my work flow. So then I tried

pd = makedist('DiscreteUniform','Lower',1,'Upper',N_clusters);
idx = random(pd,N_clusters,1);

But MATLAB gives an error that "DiscreteUniform" is not recognized. What is my best alternative usinf the random function, if I stll can.

Answer 1

[Paul]

I wonder why makedist doesn't support Uniform Distribution (Discrete).

But random can be called with that distribution.

N_clusters = 10;
rng('default');
idx = randi(N_clusters,N_clusters,1)

idx = 10×1

9 10 2 10 7 1 3 6 10 10

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

rng('default');
random('Discrete Uniform',N_clusters,N_clusters,1)

ans = 10×1

9 10 2 10 7 1 3 6 10 10

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

●

[Snoopy]

This

idx = random('Discrete Uniform',N_clusters,[N_clusters 1]);

seems to work. I wonder about the makedist too. This caused the whole confusion on my end.

Answer 2

[Matt J]

idx = floor(  (N_clusters-1)*rand(N_clusters,1) )+1

[Snoopy]

Thanks but I need to use the random function not rand.

[Matt J]

Fine but random can imitate rand() using the 'Uniform' distribution selection (which does exist, unlike 'UniformDiscrete').

[Snoopy]

Yes but I need to stick to the random function. I cannot use the rand function.

[Matt J]

Then replace the rand() function in my code with the random() function, invoked appropriately to create a uniform distribution.

[Snoopy]

Simply replacing did not the trick if I am not overlooking.

[Matt J]

It should. We would have to see what you did.

[Snoopy]

This

idx = floor((N_clusters-1)*random(N_clusters,1))+1;

gives

Error using random (line 69)
Probability distribution name must be the name of a supported distribution.

[Snoopy]

AI suggested

pd = makedist('DiscreteUniform','Lower',1,'Upper',N_clusters);
idx = random(pd,N_clusters,1);

but that also does not work.

[Torsten]

These are the distributions that are available for the "makedist" command:

list = makedist

list = 29×1 cell array
    {'beta'                   }
    {'binomial'               }
    {'birnbaumsaunders'       }
    {'burr'                   }
    {'exponential'            }
    {'extremevalue'           }
    {'gamma'                  }
    {'generalizedextremevalue'}
    {'generalizedpareto'      }
    {'halfnormal'             }
    {'inversegaussian'        }
    {'logistic'               }
    {'loglogistic'            }
    {'lognormal'              }
    {'loguniform'             }
    {'multinomial'            }
    {'nakagami'               }
    {'negativebinomial'       }
    {'normal'                 }
    {'pearson'                }
    {'piecewiselinear'        }
    {'poisson'                }
    {'rayleigh'               }
    {'rician'                 }
    {'stable'                 }
    {'tlocationscale'         }
    {'triangular'             }
    {'uniform'                }
    {'weibull'                }

[Snoopy]

Yes, that is the problem and hence I am seeking a solution to that, using still the random function.

[Matt J]

AI suggested

But that's not what I suggested to you earlier. I suggested 'Uniform'

N_clusters = 10;
rng('default');

pd = makedist('Uniform','Lower',1,'Upper',N_clusters+1);
idx = floor(random(pd,N_clusters,1))

idx = 10×1

9 10 2 10 7 1 3 6 10 10

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

●