Skip to main content

Table 2 Evaluation Measures for Implemented Models

From: Cyberbullying detection: advanced preprocessing techniques & deep learning architecture for Roman Urdu data

Evaluation Measures

Approach/Model

RNN-LSTM

RNN-biLSTM

CNN

0

1

0

1

0

1

Precision

0.90

0.72

0.88

0.70

0.84

0.60

Recall

0.91

0.68

0.91

0.64

0.90

0.46

F1-Score

0.90

0.70

0.90

0.67

0.87

0.52

Accuracy

0.85

0.84

0.79

Macro average

0.80

0.78

0.69

Weighted Average

0.85

0.84

0.78

Approximate Loss

0.6

1.2

1.7