Revisiting city tourism in the longer run: an exploratory analysis based on LBSN data

Figures & data

Figure 1. Analytical framework to interpret changes in tourism destinations based on data from LBSN.

The workflow describing a three-step process to obtain spatial and temporal signatures using LBSN data, and methods of spatial analysis for the study of city tourism.

Figure 2. Monthly distribution (ln) along the time-series for tourists (left) and tourists' photos (right). Black dots correspond to the monthly values by year, and the red dots refer to the mean values.

A set of violin plots describing the monthly distribution of Flickr–based estimates of tourists, and tourists’ photos using LBSN data between 2007 and 2017.

Table 1. Pearson correlation (r), coefficients of determination, F-test, and significance level between monthly visitor counts from Flickr and monthly hotel occupancy rates from 2012 to 2017.

Year	r	Adjusted R^2	F	F significance (Level = 0.05)
2012	0.69*	0.45**	2.43	0.16
2013	0.81**	0.67***	3.11	0.07
2014	0.86***	0.72***	2.40	0.16
2015	0.64*	0.36*	1.16	0.80
2016	0.89***	0.82***	1.96	0.27
2017	0.71**	0.59**	3.61	0.05

*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001.

Figure 3. Spatial distribution of city tourists in Lisbon based on Flickr data. The figure shows Flickr-based estimates of visitors in each specific area (grid cell) by year (2007, 2010, 2014, and 2017).

The spatial distribution of tourists in Lisbon based on LBSN data and the shifts in the spatial concentration and dispersal of city tourists for 2007, 2010, 2014, and 2017.

Figure 4. Group categories from spatiotemporal clustering.

Five groups of clusters overlapping the area of Lisbon city that describe the distinct spatial and temporal patterns of tourist visits in different tourism places across the urban space.

Table 2. Descriptive statistics of aggregated data with reference to the space–time cube.

Time-step	From	To	n° bins	Sum	Mean	Std. Dev.
1^th	01/01/2007	31/12/2007	343	773	2.3	3
2^th	01/01/2008	31/12/2008	394	926	2.4	3
3^th	01/01/2009	31/12/2009	407	1,154	2.8	4.2
4^th	01/01/2010	31/12/2010	565	1,648	2.9	5.2
5^th	01/01/2011	31/12/2011	793	2,397	3.1	5.8
6^th	01/01/2012	31/12/2012	1,047	4,092	4	8
7^th	01/01/2013	31/12/2013	1,107	4,705	4.3	8.6
8^th	01/01/2014	31/12/2014	1,045	5,037	4.9	10.1
9^th	01/01/2015	31/12/2015	1,111	5,489	5	10.5
10^th	01/01/2016	31/12/2016	986	4,674	4.8	9.6
11^th	01/01/2017	31/12/2017	1,012	4,364	4.4	8.6

Table 3. Descriptive statistics of the group categories from spatiotemporal clustering.

Category	n° hexagons	Visits
		Sum	Mean	Std. Dev.
Consecutive	133	20,500	154	144
Mid-term consecutive	101	5,649	56	46
Intensifying	37	1,275	35	14
New	47	1,820	39	23
Sporadic	81	1,592	20	5

Figure 5. Tourist flows between parish areas in Lisbon, based on movement data derived from Flickr (2010, 2013, and 2017), and Twitter (2018, and 2019). The figure shows the resulting scores from centrality measures computed for each parish area.

The changes in the spatial distribution of tourist flows between parish areas in Lisbon, using LBSN data derived from Flickr and Twitter for the years 2010, 2013, 2017, 2018, and 2019. The flow patterns suggest the homogenization of tourist flows in urban destinations.