MATLAB and Simulink resources for Arduino, LEGO, and Raspberry Pi

Learn moreOpportunities for recent engineering grads.

Apply Today**New to MATLAB?**

Asked by Leif
on 1 Jun 2013

I have two large vectors. The first is a list of measurement values and the second is a categorical identification variable (5 different variables). I am trying to create a matrix of 5 columns (one for each categorical identifier) with all of the measurement values as the different rows?

Thanks

*No products are associated with this question.*

Answer by the cyclist
on 1 Jun 2013

Edited by the cyclist
on 1 Jun 2013

If you are guaranteed the same number of categorical variable, then

[~,idx] = sort(categoricalVariable); measurementVariable = measurementVariable(idx); measurementVariable = reshape(measurementVariable,[],5);

Leif
on 1 Jun 2013

Unfortunately no I do not have the same number of categorical variables. The numbers are the results of a particle tracking model so the variability in the number of values is very high.

Image Analyst
on 1 Jun 2013

What are the categories? Do you have multiple types of particles, so you'd have something category #1 = "particle of type #1", and category #2 = "particle of type #2", and so on? Hopefully you're using standard terminology as outlined in http://en.wikipedia.org/wiki/Categorical_variable, because I don't really see why the number of categories should change, even if the number of particles changes. A category could always have a count of zero if no particles of that category existed in that movie frame. Posting an image or figure or diagram would probably help me to understand.

Answer by Andrei Bobrov
on 1 Jun 2013

mv = randi(120,150,1); %measurement values cv = randi([1 5],150,1);%categorical identification variable

out = accumarray(cv,mv,[],@(x){x});

## 2 Comments

## the cyclist (view profile)

Direct link to this comment:http://www.mathworks.com/matlabcentral/answers/77656#comment_152299

Are you guaranteed that the categorical identification variables appear the same number of times, so that your columns have equal length? Or might you need to pad them?

## Image Analyst (view profile)

Direct link to this comment:http://www.mathworks.com/matlabcentral/answers/77656#comment_152300

How is the second vector "large" when it has only 5 categorical identifications, like {'Class1', 'Class2', 'Class3', 'Class4', 'Class5'}? Or is the second vector really a N by 5 array of numbers??? How large is "large"? 50 million? 300 million? If you're talking about something less than a few million, that's not large. If only the

firstvector is measurements, how are you going to get 5 more columns of measurements for the output array? Where are those extra measurements going to come from or how are they to be calculated? I think you'd best give a small example, maybe with 5 or 10 rows or something.