COVID-19 Infodemic in Malaysia: Conceptualizing Fake News for Detection

Table 5

Summarized results of algorithm models of COVID-19 fake news detection.

Algorithm	Model	Train score (%)	Test score (%)	Accuracy (%)	Precision (%)	Recall (%)	F1-score (%)
Logistic regression	Baseline	76.37	72.44	72.44	72	100	84
	Before SMOTE	99.18	73.08	73.08	74	96	84
	After SMOTE	100	72.44	72.44	73	98	84
Naïve Bayes	Baseline	83.24	72.44	72.44	72	100	84
	Before SMOTE	83.24	72.44	72.44	72	100	84
	After SMOTE	100	73.08	73.08	78	88	82
Decision tree	Baseline	100	71.79	71.79	76	88	82
	Before SMOTE	77.75	72.44	72.44	72	100	84
	After SMOTE	94.60	73.72	73.72	79	88	83
Support vector machine	Baseline	100	72.44	72.44	72	100	84
	Before SMOTE	100	72.44	72.44	73	98	84
	After SMOTE	100	72.44	72.44	73	98	84
Random forest classifier	Baseline	100	73.08	73.08	73	100	84
	Before SMOTE	84.89	72.44	72.44	74	96	84
	After SMOTE	89.56	73.08	73.08	74	98	84
Gradient boosting classifier	Baseline	93.13	72.44	72.44	74	96	83
	Before SMOTE	88.19	74.36	74.36	75	97	85
	After SMOTE	98.38	75.64	75.64	80	89	84

The most suitable machine learning algorithm model is bold.