The testing set in data partitioning is the:

first subset of data, which usually contains $30\%$ of the records initial dataset from which subsets are created second subset of data, which usually contains $30\%$ or less of the records first subset of data, which usually contains $70\%$ of the records