Temporal and spatiotemporal investigation of tourist attraction visit sentiment on Twitter
Linear regression models are fit according to the average sentiment score at attraction visits vs. visit number in the day. The first graph in the first row is gathered from the entire user set whereas the following three graphs are based on different visitor origins. Shaded areas around the linear regression line contains the 95% confidence interval and the round markers in each visit number marks the average sentiment score for that visit. We note that all regression fits are statistically significant.