This document evaluates four unsupervised feature selection methods for positive-unlabeled learning in text classification. It compares Collection Frequency (CF), Document Frequency (DF), Collection Frequency-Inverse Document Frequency (CF-IDF), and Term Frequency-Document Frequency (TF-DF). The document collected a dataset of positive diabetes webpages and unlabeled webpages to test the feature selection methods. It found that the DF (Document Frequency) method, which selects features based on the number of documents a feature appears in, was the most effective at feature selection for positive-unlabeled learning based on the experiments.