## Sorting by multiple columns in a table with natural number sorting

### Lilly Aurelius (view profile)

on 20 Apr 2019
Latest activity Commented on by Cris LaPierre

on 21 Apr 2019

### Cris LaPierre (view profile)

I have a table of data that I want to sort by pair number and by trial number.
The issue is that if I use:
data = sortrows(data, {'pairNum', 'trialNum'}, {'ascend', 'ascend'});
It sorts with trial 10 being before trial 1.
However, if I try and use Stephen Cobeldick's nat sort function, I can't take into account the pair number as well as the trial number so it ends up sorting with all trial 1s, then all trial 2s etc.. (doesnt preserve the order of the pairs).
Is there a way to sort my table so that my pairs go from 1-100, and within those pairs, the trial numbers go from 1-100?
Also, I tried adding zeros to the trial numbers so that it would hopefully sort 01 before 10, but when I convert to a number using str2num and/or str2double, it gets rid of the leading zeros.
data.trialNum = strsplit(sprintf('%02d ',data.trialNum))';
The datatset is attached as a .mat file.

Andrei Bobrov

on 20 Apr 2019
Lilly Aurelius

### Lilly Aurelius (view profile)

on 20 Apr 2019
Thanks, I've attached a sample.

R2017b

### Cris LaPierre (view profile)

on 20 Apr 2019
Edited by Cris LaPierre

### Cris LaPierre (view profile)

on 20 Apr 2019

Here's a simple example that works for me:
pairNum = randi(15,1,200)';
trialNum = randi(15,1,200)';
data=table(pairNum, trialNum)
data = sortrows(data,{'pairNum','trialNum'})

Cris LaPierre

### Cris LaPierre (view profile)

on 20 Apr 2019
Sorting ascended is default, so you don't have to specify it.
The data set you shared does not contain a trialNum == 1. You also talk about having to use str2num (better to use str2double). Could you include a raw data set? Even when I convert the numbers to strings, I don't observe the behavior you describe.
How are you loading your data into MATLAB? If using readtable or the import tool, you can specify the column data type, which means you can import your numbers as numbers, not strings, bypassing this whole issue.
Lilly Aurelius

### Lilly Aurelius (view profile)

on 21 Apr 2019
Thanks. I ran my code line by line and you're right, it seems to be working. I must have been doing something wrong...
The trial numbers are orignially strings ('Trial1', 'Trial2', etc) so I convert the cell array of strings into a character array, and then extract the numbers at the end of each string and finally convert to numbers.
Cris LaPierre

### Cris LaPierre (view profile)

on 21 Apr 2019
There is also a categorical data type. That may work for your needs. Check it out.