SimilarityMeasure initialization problem


HI all,
I have a dataset with an id column of type "text" named "att24754".
I have a problem with initialization of Cosine similarity measure. Calling the following code:
"The example set contains non-numerical attribute att24754 which is not allowed for value based similarities."
and the strange thing is that "att24754" has set to the "id" of my exampleset. So I expect it to not be involved in similarity calculation as an "special attribute". But it will be involved in the similarity computation and so RM generates the error. I know all of the attributes should be numerical except my id. Does anyone know what's going wrong here?
Any idea would be greatly appreciated.
thanks.
I have a dataset with an id column of type "text" named "att24754".
I have a problem with initialization of Cosine similarity measure. Calling the following code:
Now the problem is come up in calling init "method":
clusteredSet=clusteredSetInput.getData(ExampleSet.class);
similarityMeasure= new CosineSimilarity();
similarityMeasure.init(clusteredSet);
"The example set contains non-numerical attribute att24754 which is not allowed for value based similarities."
and the strange thing is that "att24754" has set to the "id" of my exampleset. So I expect it to not be involved in similarity calculation as an "special attribute". But it will be involved in the similarity computation and so RM generates the error. I know all of the attributes should be numerical except my id. Does anyone know what's going wrong here?
Any idea would be greatly appreciated.
thanks.
Tagged:
0
Answers
You may also check your attributes in your source code. Basically the init method iterates over the attributes object of the example set which returns only regular attributes.