Classifier details

Basic data
Classifier ID27
TitleSentiment (Product reviews + Tweets)
DescriptionThis classifier has been trained with tweets and product reviews classified by sentiment (positive or negative). Use it to classify texts of unknown nature.
Created2012-01-18 12:04:36 MET
Distinct texts85536
Precision (biased) 76.3585% (sample standard deviation: 0.529586%)
Precision (unbiased) 76.6706% (sample standard deviation: 0.529931%)
Precision has been measured with 10-fold cross-validation

Training class distribution

Precision per class

Class Precision (biased) Precision (unbiased) % of mistakes (biased) % of mistakes (unbiased)
positive 91.0419% 89.9976% 79.6805% 77.0083%
negative 59.3793% 61.26% 20.3195% 22.9917%

(c) 2017 netEstate GmbH • Imprint