In this blog post graphic methods are used to describe tourism time series for six independent states of former Yugoslavia. The following six graphs display each series on a separate panel. On such graphs any persistent pattern related to trend or seasonal factors should be clearly visible because each graph has its own vertical scale.
Figure 1 shows line graphs of total tourists arrival in six independent states.
Graphs in Figure 1 could be used to visually assess similarity in the six time series patterns. The first impression is that the total tourists arrival in Croatia and Montenegro have quite similar pattern. Then the Slovenia and Macedonia total tourists arrival time series could be added to this group. Time series for Bosnia and Herzegovina and Serbia would probably be in their own group.
In the future blog posts cluster method will be used to classify these time series into similar clusters using appropriate time series distance measures such as ARIMA model for each series or autocorrelation function.
Figure 2 shows line graphs of domestic tourists arrival in six independent states.
It is difficult to decide on the similarity of the domestic tourists arrival time series in these six independent states. All series look quite distinct, though the Croatia and Montenegro time series indicates that they might be in the same cluster.
Figure 3 shows line graphs of foreign tourists arrival in six independent states.
Visually inspecting graphs in Figure 3 lead us to similar conclusion we have reached when inspecting total tourists arrival time series for these six independent states. The first impression is that the foreign tourists arrival in Croatia and Montenegro have quite similar pattern. Then the Slovenian foreign tourists arrival time series could be added to this group. Bosnia and Herzegovina, Macedonia and Serbia would probably be in their own group.
This visual classification will be formally verified later in the future blog posts by using time series cluster methods.
Figure 4 shows line graphs of total tourists overnight stay in six independent states.
Graphs in Figure 4 indicate that the total tourists overnight stay in Croatia and Montenegro have quite similar pattern. Then the Slovenia and Macedonia total tourists overnight stay time series could be added to this group. Time series for Bosnia and Herzegovina and Serbia would probably be in their own group.
Figure 5 shows line graphs of domestic tourists overnight stay in six independent states.
As with the domestic tourists arrival time series it is difficult to decide on the similarity of the domestic overnight stay time series in these six independent states. All series look quite distinct, though the patterns of Croatia and Montenegro time series indicate that they might be in the same cluster, with Macedonian time series added later to the same cluster.
Figure 6 shows line graphs of foreign tourists overnight stay in six independent states.
Visually inspecting graphs in Figure 6 lead us to similar conclusion we have reached when inspecting foreign tourists arrival time series for these six independent states. The first impression is that the foreign tourists overnight stay in Croatia and Montenegro have quite similar pattern. Then the Slovenian foreign tourists overnight stay time series could be added to this group. Bosnia and Herzegovina, Macedonia and Serbia would probably be in their own group.
This visual classification will be formally verified later in the future blog posts by using time series cluster methods.
