I have a character vector with repeated elements and want to extract each element once and at the sequence of appearance.

Hi all,

I have a character vector with repeated elements and want to extract each element once, at the sequence of appearance in the original file. I am attaching an example file. In fact, I need to get:

Group Frontal_L Ins_Cing_L etc..

I guess I will have to use regexp but not very experienced with this functionality. Any ideas very welcome.

Answer 1

[dpb]

[unames,ia,iu]=unique(strtrim(cellstr(Getcurvenames)),'stable');

unames will be the names that are unique in the overall list,

unames=Getcurvenames(ia);  % ia is location vector in original list Getcurvenames
Getcurvenames=unames(iu);  % rebuild the original from the indices in unique list

unames will be cellstr list of names, much more conveniently handled than are fixed-length char arrays.

ADDENDUM

I forgot about the request for initial order in the output; use the 'stable' option as shown above. To not return the group string, make a slight modification to eliminate it first --

cnames=strtrim(cellstr(Getcurvenames));     % convert to trimmed cellstr
cnames=cnames(~contains(cnames,'Group'));   % keep only those ~='Group'
[unames,ia,iu]=unique(cnames)),'stable');   % and the unique list of those

[GioPapas81]

Thank you very much dpb, but I get a variable with only the initial letter of each name, which is still repeating (see attached).

[GioPapas81]

Just for the history, I found a way to do that, slightly modifying your answer dpb:

[~,idx]=unique(strtrim(cellstr(Getcurvenames)));

% Sort by ascending order the indices idx=sort(idx);

% Convert to char array to a cell array NGetcurvenames=cellstr(Getcurvenames);

% Extract only the indexed arrays starting from zero to exclude the % unnecessary 'Group' curves. for i=2:length(idx);Extractcurvenames{i}=NGetcurvenames{idx(i)};end Extractcurvenames=Extractcurvenames(~cellfun('isempty',Extractcurvenames));

It now works! Thank you again!

[dpb]

Don't know what you did differently, works fine here--

>> [u,ia,iu]=unique(strtrim(cellstr(Getcurvenames)));
>> u
u =
  15×1 cell array
    {'Central_L'    }
    {'Central_R'    }
    {'Frontal_L'    }
    {'Frontal_R'    }
    {'Group'        }
    {'In Mask'      }
    {'Ins_Cing_L'   }
    {'Ins_Cing_R'   }
    {'Occipital_L'  }
    {'Occipital_R'  }
    {'Parietal_L'   }
    {'Parietal_R'   }
    {'Posterior_all'}
    {'Temporal_L'   }
    {'Temporal_R'   }
>>

Returning a single character is characteristic of still dealing with char() arrays that are just 2D arrays of bytes and not using the trailing : to address the full array; hence the conversion to cellstr() that addresses the full cell.

unique does return the unique values in sorted order by default, there is the 'stable' optional order input to return in discovered order if that's what's wanted--

    >> [u,ia,iu]=unique(strtrim(cellstr(Getcurvenames)),'stable');
  >> u
  u =
    15×1 cell array
      {'Group'        }
      {'Frontal_L'    }
      {'Ins_Cing_L'   }
      {'Temporal_L'   }
      {'Occipital_L'  }
      {'Parietal_L'   }
      {'Central_L'    }
      {'Frontal_R'    }
      {'Ins_Cing_R'   }
      {'Temporal_R'   }
      {'Occipital_R'  }
      {'Parietal_R'   }
      {'Central_R'    }
      {'Posterior_all'}
      {'In Mask'      }
  >>