Research Article

Gamma-Poisson Distribution Model for Text Categorization

Table 2

Classification accuracy on the 20 Newsgroups dataset with two different methods for estimating parameters for the gamma-Poisson distribution. Values are shown as accuracy where is the standard deviation calculated through 10-fold cross-validation.

Feature selection Accuracy
Rational approximation Iterative method

Initial vocabulary
CF ≥ 5
CF ≥ 10
CF ≥ 20
CF ≥ 50
CF ≥ 100