MATLAB: Do I obtain an error about string cells when I try to join two datasets in MATLAB (R2013a)

datasetsjoin;Statistics and Machine Learning Toolbox

I am using the function JOIN on two datasets:

>> C = join(dataset1,dataset2);

but I obtain the following error:

 Error using dataset/join (line 274)
 Left and right key variables 'key1' and 'key1' include cells containing non-string values.

Best Answer

The reason behind the error is that the content of your dataset, under the variable name 'key1' must be of type 'char", and not "double", as in your case.

To circumvent this issue, you need to do a data-type conversion. Here is an example that illustrates that. Say, your variable name tagged "key1" is located in column "idx1" in "dataset1" and in column "idx2" in "dataset2".

>> varName1 = dataset1.Properties.VarNames(idx1);
>> cellData1  = dataset2cell(dataset1(:,idx1));
>> data1    = cellData1(2:end,2);
>> data1    = cellfun(@num2str,data1,'UniformOutput',false);
>> ds1      = dataset(data1,'Varnames',varName1,'ObsNames',dataset1.Properties.ObsNames);
>> dataset1(:,idx1) = ds1;
>> varName2 = StatSet.Properties.VarNames(idx2);
>> cellData2  = dataset2cell(dataset2(:,idx2));
>> data2    = cellData2(2:end,2);
>> data2    = cellfun(@num2str,data2,'UniformOutput',false);
>> ds2      = dataset(data2,'Varnames',varName2,'ObsNames',dataset2.Properties.ObsNames);
dataset2(:,idx2) = ds2;
C = join(dataset1,dataset2,'Key','key1');

Related Solutions

MATLAB: Intersection between a dataset and a cell array

Okay, I've never worked with datasets but they are quite similar to tables.

%%Load data
d1=load('xyz.mat');
d2=load('abc.mat');
%%Extract cell array with xyz names
xyzNames=d1.ObsNames_ds_ACP_selected_vars
%%Extract cell array with abc names and values
abc=d2.ds_ACP_selected_vars
abcNames=abs.Properties.ObsNames
abcValues=abc.Classvar;
%%Compare
[~,b]=intersect(abcNames,xyzNames)
%%Names
abcNames(b)

ans =

4×1 cell array
    {'CCImpOpA'          }
    {'ProdottiAccesiMens'}
    {'ProdottiInEssere'  }
    {'classe_eta'        }
%%corresponding values
abcValues(b)

ans =

MATLAB: How to change the variable or observation names in the DATASET object in MATLAB 7.9 (R2009b)

DATASET objects have SET and GET methods defined. Please see the documentation for DATASET SET and GET methods for more information on usage.

Please also note that these methods are similar to the SET and GET methods defined for handle graphics, except for one important difference: to make a change on object OBJ, the object output from a SET command on OBJ must be reassigned to OBJ. If this reassignment is not done, the changes will not be made.

An example of changing the variable names of a dataset object for the IRIS object created above is as follows:

iris = set(iris,'VarNames',{'varname1','varname2','varname3','varname4','varname5'})

As per the reassignment discussion above, if we had executed the following, the changes would not have been made:

set(iris,'VarNames',{'varname1','varname2','varname3','varname4','varname5'})

To change only a single variable name, execute:

varnames = get(iris, 'VarNames');
varnames{3} = 'MyVarName';
iris = set(iris, 'VarNames', varnames);

You can also change a single variable name by explicitly setting this property, i.e.:

iris.Properties.VarNames{1}='newName'

Best Answer

Related Solutions

MATLAB: Intersection between a dataset and a cell array

MATLAB: How to change the variable or observation names in the DATASET object in MATLAB 7.9 (R2009b)

Related Question