Benchmark Surface
Model Evaluation
Compare supported sentiment models across available corpora, inspect confusion structure, and review class-level behavior before using them in the public dashboard.
Dataset
Corpus
Switch between evaluation corpora without leaving the page or changing chart semantics.
Selection
Model
Choose a single model to drive the confusion matrix and per-class visual diagnostics.
Snapshot
Summary
Key model quality metrics for the currently selected corpus and evaluation target.
Dataset: Train5 Corpus
ModelNaive BayesSelected evaluation target
Accuracy74.1%Overall classification accuracy
Precision73.2%Macro average
Recall72.0%Macro average
F172.1%Macro average
Visuals
Evaluation Visuals
Read prediction concentration and per-class tradeoffs side-by-side for the active model.
Reference Table
All Model Details
A full comparison table for the loaded corpus, including confusion matrix shape.
| Model | Accuracy | Precision (macro) | Recall (macro) | F1 (macro) | Confusion Matrix Shape |
|---|---|---|---|---|---|
| Naive Bayes | 74.1% | 73.2% | 72.0% | 72.1% | 3 x 3 |
| SVM | 78.4% | 76.6% | 76.2% | 76.4% | 3 x 3 |
| VADER | 74.0% | 72.6% | 71.3% | 71.4% | 3 x 3 |