Attribute Selection Hybrid Network Model for risk factors analysis of postpartum depression using Social media

Brain Informatics

Table 5 Results of evaluation on the test set

Method	Precision	Recall	Accuracy	F1 score
BOW^a	51.68±9.89	52.17±3.70	64.59±3.79	53.84±6.35
BOW^a	55.43±1.99	59.82±1.88	61.12±4.46	62.92±1.51
GloVe^a	58.21±5.52	56.30±7.20	58.07±6.33	50.66±5.80
GloVe^b	60.64±6.67	62.47±6.06	63.18±2.68	60.12±3.16
Feature rich^a	59.80±6.21	59.80±6.21	70.91±6.81	54.47±3.66
Feature rich^b	62.60±7.77	60.26±7.88	74.89±4.05	61.59±2.20
Baseline	70.78±6.07	73.74±11.24	69.29±2.64	75.12±2.76
ASHN	78.77±2.99	72.94±1.88	74.77±3.24	75.74±1.33

^a and ^bindicate MNB and SVM classifiers, respectively
Bolder values indicate the better results than other attribute selection methods