Table 4.

Top reported system performance for studies predicting the age of Twitter users using traditional machine learning (ML) methods. Result metrics are reflected in this table as reported in the original publications and are not directly comparable to each other. Reviews are ordered by the number of classification groups.

Study	Number of age groups	Age class detail (y)	Language	ML method	Reported performance
					F1-score	Accuracy
Jurgens et al [80], 2017	N/Aa	Continuous	English	RFb regression	N/A	0.71
Volkova [110], 2017	2	18-23 and 25-30	English and Spanish	LogRc	N/A	0.77
Xiang et al [116], 2017	2	≤30 and >30	English	CPMEd	N/A	0.74
Ardehaly and Culotta [53], 2018	2	<25 and >25	English	LLPe	N/A	0.78
Morgan-Lopez et al [90], 2017	3	13-17, 18-24, and >24	English	LogR	0.74	N/A
Arafat et al [51], 2020	3	≤24, 25-39, and ≥40	NRf	LogR	N/A	0.71
Cornelisse and Pillai [66], 2020	3	18-24, 25-54, and >55	English	LogR	0.78	N/A
Markov et al [87], 2017	5	18-24, 25-34, 35-49, 50-64, and >65	English, Spanish, Dutch, and Italian	LogR	N/A	0.56-0.65
Cheng et al [65], 2018	5	18-24, 25-34, 35-44, 45-54, and 55-64	English, Filipino, and Taglish	SVCg	0.61	0.86
Garcia-Guzman et al [70], 2020	4	18-24, 25-34, 35-49, and >50	English	Bag of trees	N/A	0.67
Chamberlain et al [64], 2017	10 (3 subgroups)	<12, 12-13, 14-15, 16-17, 18-24, 25-34, 35-44, 45-54, 55-64, and >64	English, Spanish, French, and Portuguese	Bayesian probability	0.31-0.86 (3 class)	N/A

^aN/A: not applicable.

^bRF: random forest.

^cLogR: logistic regression.

^dCPME: coupled projection matrix extraction.

^eLLP: learning with label proportions.

^fNR: not reported.

^gSVC: support vector classifier.