Table 1.
The OSM data in the areas used in our study.
Fig 1.
10-fold cross validation F-score of the six different classifiers for the eight basic emotion types.
Fig 2.
Performance metrics for the models used in the study (10-fold cross validation).
Table 2.
Examples of tweets classified for each emotion in London and San Francisco.
Fig 3.
A time based heatmap of San Francisco (red = high levels of emotion, blue = low levels of emotion).
Fig 4.
A time based heatmap of London (red = high levels of emotion, blue = low levels of emotion).
Fig 5.
The emotions displayed on different days of the week in San Francisco.
Fig 6.
The emotions displayed on different days of the week in London.
Table 3.
The top 10 POI for each emotion for the cities of San Francisco and London.
Fig 7.
The emotions displayed for different location categories in San Francisco (based on the mean rank. A higher rank denotes a higher level of emotion).
Fig 8.
The emotions displayed for different location categories in London (based on the mean rank. A higher rank denotes a higher level of emotion).
Table 4.
Results of significantly different categories: (1): Hotel & Restaurants, (2): Commercial Services, (3): Attractions, (4): Sports & Entertainment, (5): Education & Health, (6): Public Infrast., (7): Manufacturing & Production, (9): Retail, (10): Transport, (11): Residential, (12): Office.
Table 5.
The categories of locations which showed significant differences (p<0.05) in the display of emotions at three different distance levels in San Francisco.