Table 3 Model accuracy on an external dataset.

From: Twitter Sentiment Geographical Index Dataset

Dataset

Size

Share in 2021 tweet

Accuracy

Precision

Albanian

22,126

—

69.2

83.6

Bosnian

13,621

—

74.7

74.6

Bulgarian

12,320

0.02%

72.1

73.1

Croatian

45,505

—

79.6

82.3

English

22,844

38.66%

78.6

77.0

German

29,705

0.77%

74.5

73.9

Hungarian

26,880

0.06%

75.5

89.5

Polish

84,758

0.41%

71.7

73.7

Portuguese

34,539

13.45%

57.9

48.6

Russian

23,751

0.82%

71.3

67.0

Serbian

21,311

0.04%

63.4

53.2

Slovakian

37,021

—

77.2

80.7

Slovenian

42,978

0.05%

72.4

66.7

Spanish

81,143

12.42%

67.9

86.1

Swedish

20,068

0.20%

67.7

58.4