quickSentiment 0.3.2

Major Changes & API Updates

• pipeline() no longer computes ROC or returns ROC guidelines
• predict_sentiment() no longer inherits a threshold directly from the pipeline. It now defaults to a standard 0.5 classification threshold, while still allowing users to manually pass the custom cutoff they feel most comfortable with.
• evaluate_performance() is now available. Users can now pass model predictions and test data to calculate detailed metrics (including AUC, Precision, and Recall) targeted at any specific factor level or class of interest.

Performance & Dependency Improvements

• pipeline() does less computing then before

New Features

• The new evaluate_performance() engine leverages R’s S3 object-oriented system. It returns custom evaluation objects that natively integrate with base R’s plot() function to instantly generate high-quality ROC and Precision-Recall curves.