Consequently, the new baseline likelihood of the term-founded classifier so you’re able to categorize a profile text message on best relationships classification is actually fifty%
To do this, 1,614 texts of every relationships group were utilized: the whole subset of your own set of casual matchmaking seekers’ texts and you will an equally large subset of your ten,696 messages to your enough time-label relationships seekers
The phrase-founded classifier is dependant on the classifier approach out-of Van der Lee and you will Van den Bosch (2017) (see along with Aggarwal and you may Zhai, 2012). Half dozen different server reading tips are used: linear SVM (assistance vector servers), Unsuspecting Bayes, and five alternatives of tree-founded formulas (choice tree, haphazard tree, AdaBoost, and you may XGBoost).