Possibilities of using Selected Visualization Methods for Historical Analysis of Sporting Event – An Example of Stage Cycling Race Tour de France

The final map development was preceded by an extensive process covering the processing of historical data from particular Tour de France editions. For the purpose of efficient data handling, as well as due to the attempt to distribute the obtained results on the Internet in the form of a website dedicated to the Tour de France

The website dedicated to the Tour de France prepared by the authors of the paper, is available at: www.tdfrance.eu.

Fig. 1

Database tables were designed to allow effective handling of these data on the created website. An important consideration in the design was the fact that these attributes saved in tables must allow answering any questions related to the history of the race under review. The standard SELECT command, based on which we obtain the final presentation of necessary data, was used for particular queries. A basic example of the use for practical questions can be found in Table 1.

The website (tdfrance.eu) is designed so that a majority of answers to these questions is available to clients who are interested in such information. Certain functions and attributes are available to the administrators only; however, their implementation is only a question of the presentation of the function on the specific subpage.

The chord diagrams were used to illustrate the quantity of achievements of relevant cyclists or nations in a specific stage or mountain ascent on a main map document. In this case, data from the database were stored (after the relevant selection taking into account the necessary criteria) as simple *.txt files. The diagrams were developed using the Circos project, the author of which is Martin Krzywinski (Genome Sciences Centre, Vancouver). The primary aim of the Circos project was circular genome visualization, however, such an interpretation also has a wide-ranging application in other scientific disciplines. This form of data processing can be interpreted easily and intuitively and, in combination with the map visualization, constitutes an interesting tool for the presentation of a high quantity of not only genome data but also spatial data. Diagrams generated through the online service

The service is available at: http://mkweb.bcgsc.ca/tableviewer/.

The internal structure of a cyclist's victories (and nationalities) in each finish places and mountain ascents, is captured by using a circular visualization, in the form of chord diagrams. This form of presentation is interesting and allows the display of large amounts of data. Data from the database (MySQL tables) were adjusted to the required form using standard SQL commands. The resulting datasets were transformed into the output format (*.txt file in the case of an offline chord diagram on the map, *.json file in the case of interactive online presentations). D3 JavaScript library was also used in the online presentation as well as on the online maps (see section D3 and TopoJSON). The result is a circular visualization of selected historical data from Tour de France (Fig. 2).

Fig. 2

Circular visualization is an useful tool for a displaying large amount of data. The interactive online environment allows the client to read a specific database using a variety of scripting languages (e.g. JavaScript). In offline mode, the client must properly interpret the displayed data. Basic interpretation of this circular visualization is shown in a chord diagram showing the most successful cyclists in the history of the Tour de France, according to the number of winning stages (Fig. 3).

Fig. 3

The outer circumference of diagram is formed by individual matrix entities (cyclists and stage finishes, where the selected cyclists triumphed). Cyclists are ordered by the number of winning stages. Finish places are divided into four basic categories (see Fig. 3 legend). The base width of the relevant cyclist (number 1 in Fig. 3) corresponds to the number of stage triumphs (e. g. Eddy Merckx – 34 stage triumphs). Under the base are small boxes of different colours, with the colour corresponding to the type of finish place in which the cyclist triumphed and the thickness expresses the number of stage wins at a given location. The cyclist and the finish place are connected with a ribbon of appropriate width (number 3 in Fig. 5). In the case of a finish place, the width of the base expresses the number of the observed cyclist’s triumphs at this place. The individual segments reflect the triumphs of concrete cyclists (number 2 in Fig. 5). In this example we can see that the observed cyclists reached altogether 14 stage victories in Paris (Eddy Merckx and Mark Cavendish – 4 victories, André Leducq – 3, Bernard Hinault – 2 and André Darrigade – 1 triumph). The city of Paris is the only common place where all of the above cyclists triumphed (Merckx – 4 victories, Cavendish – 4, Leducq – 3, Hinault – 2, Darrigade – 1). Here, however, we can say that it is just the Paris triumph that is one of the most valuable victories for cyclists participating in the Tour de France. At this point, it is appropriate to draw attention to the basic principle of the final cyclist's selection. There was only one criterion of this selection and it was the total number of stage triumphs. Based on data organized in a database, it would be easy to add similar characteristics to any group of cyclists. For example, five and three-time Tour de France champions (Eddy Merckx, Bernard Hinault, Jacques Anquetil, Miguel Indurain, Louison Bobet, Philippe Thys and Greg Lemond) triumphed altogether in 119 stages at 82 different places. In addition to basic quantitative characteristics of chord diagram we can also indirectly reveal the specific nature of these cyclists’ racing techniques. Mark Cavendish (30 stage victories) and André Darrigade (22) are typical representatives of sprinters. On the contrary, the triumphs of Eddy Merckx (34), Bernard Hinault (28) and André Leducq (25) demonstrate their universality and abilities to win even demanding mountain stages. A specific feature is the triumphs of Bernard Hinault who gained many of his wins in time trials and in initial prologues (20).

Circular visualization of these historical data represents an attractive form of data representation and its interactive design on Internet allows to the client direct identification of values. Integral part of the resulting map are four chord diagrams, while in the legend is also a reference to the online version of each chord diagram.

D3 (Data Driven Documents) is a JavaScript library for producing data visualizations of extensive amounts of data. D3 allows direct control of data while using the standard model of an object, the so-called DOM (Document Object Model). Developers can link input data to any elements of a document using D3, apply dynamic transformation, and modify the content (Bostock et al. 2011). The TopoJSON format is a text file that encodes the topology between particular map objects. At present, the universal nature of the D3 library in a combination with the presentation of spatial data in the TopoJSON format, can be considered as an effective and user-friendly method of cartographic visualizations of data. The previous section of the paper mentions the D3 library with respect to the online version of circular data visualization. This library was also used jointly with the TopoJSON format for an online presentation of maps (final points, mountain ascents). Such an online presentation has several benefits in comparison with a static map. We consider easy identification of selected places (final points, mountain ascents), and the linking of these objects to the database with relevant attributes, as the greatest benefits. Thus the client can easily display relevant data related to the selected place, and identify historical statistics. For the presentation of such a map, we must integrate basic libraries (topojson.js, d3.js) into the source file, while the libraries contain the complete functionality required for the display of the map (or of another object) on the part of the client. In addition to the above libraries, it is necessary to download the source data required for the map. The presentation of the map and the definition of variables corresponds to the standards of JavaScript. A model presentation of such a code is as follows

A detailed record of source code including defined variables can be shown directly on the website http://www.tdfrance.eu/json_finish.php.

The supporting map of the countries in the world is in the format *.topojson (world.topojson). Final points with relevant attributes are saved in *.csv file (finishes.csv). This contains several important attributes. The latitude (lat) and longitude (lon) defines the exact position of the relevant final point on the map. The attribute “ID” is the basic identifier of the given place which can show detailed statistics of the place as this value is identical to the attribute “id_place” in the database table “tdf_places” (see the section Data processing). A model final presentation of such a map is on Figure 4.

Fig. 4

The basis for creating the final map was obtaining and efficiently redistributing all data. All necessary data were transformed into map layers (*.shp format). This map, consisting of four partial maps and diagrams, we have attempted to present the most important sport achievements of individuals, including the countries in terms of stage triumphs and victories in mountain crossings, and also to illustrate the places of such historical moments. Such cartographic processing can be viewed as innovative in terms of the perception of Tour de France history and could be inspiring also for other analyses covering the history of sports in all sectors where – as for the Tour de France – space plays an important role.

On the final map (Fig. 5) we intended to visualize the main spatial-historical characteristics of the Tour de France, while showing also the most successful cyclists in history as well as the most successful countries. Therefore the key map layout (designed for A0 format) is divided into two sections (4 maps, including relevant chord diagrams). The left section presents stage triumphs; the right section presents triumphs in mountain ascents. The upper parts of particular sections show the places of stage triumphs of the most successful cyclists in terms of the number of stage triumphs, or winning crossings through mountain ascents of the second, first and “hors catégorie”. The bottom parts present the victories of relevant countries, in which they show all the places of stage finishes or of mountain ascents in the history of the race (1903–2016). However, the most important stage finishes (20 and more stage finishes) and mountain ascents (20 and more crossings) are highlighted. The symbols of these places are colour-coded by the quantity of stage triumphs or of triumphs in the mountain ascents of the most successful countries in the history of Tour de France (for stages it was France, Belgium, Italy, the Netherlands and Spain – all countries with more than 100 winning stages; for mountain ascents it was France, Spain, Belgium and Italy – all countries with more than 200 winning crossings in mountain ascents). To present an internal structure of triumphs, we used a non-traditional, but in our opinion, an adequate interpretation using chord diagrams. The figures presenting places in the diagram and on the map are identical, and thus the number of triumphs of the countries under review (cyclists) in the relevant place (the map) as well as the share of wins of particular countries (cyclists) in the relevant place (the chord diagram), is easily identifiable. The size of diagrams is expressed on all maps based on the total number of stage finishes or of mountain ascents. All of the maps applied the Robinson projection; the scales of the main maps of stage triumphs are 1: 4 200 000; for mountain ascents, the scale was 1: 2 700 000. Since the numerical or verbal identification of particular places on the maps would result in a lack of clarity of such maps, at www.tdfrance.eu we developed an interactive map presentation showing the maps. These are linked to the database, so also an exact identification of the triumphs of particular cyclists in these places is possible. An online version of the corresponding chord diagrams which are part of the map (the hyperlink is in the map) was also created. However, the maps have been designed to visualize key historical achievements (most successful nations and cyclists). The final map showed several historically relevant statistics. For the achievements of particular racers in stages or in mountain ascents, part of the legend is a list of all triumphs of such cyclists shown on the map. For the wins of particular countries, it is possible to identify directly on the map only stages with 20 and more stage finishes (or ascents with 20 and more crossings) in the whole history. However, the map shows all stages and mountain ascents included in the race itinerary at least once throughout history. For the purpose of identifying particular places and mountain ascents, an interactive map was inserted in the above stated website. These data are linked to the MySQL database, which keeps all collected attributes, and in addition to the identification of the place, it is also possible to get a detailed presentation of stages ending in the relevant stage place.

Fig. 5

The database and the visualization methods described in the preceding part of the paper were designed and used in such a manner that it is possible to identify the basic historical background of the history of Tour de France from its origination to the present when using them. Therefore, in this chapter we would like to stress certain interesting facts related to the development of the quantity of stage centres. They represent the above stated spatial dimension of the race whereas the inclusion of particular places in the itinerary of particular years significantly affects the overall character of the race. In the more than 100-year history of Tour de France we can identify a total of 682 places, which gained the status of stage places (either as a start or as a finish). 569 places are French, and the remaining places are stage places in the surrounding countries (most of them in Belgium – 39). In the first Tour de France editions, stages repeated. The route was almost identical, and new centres were included in the itinerary relatively sporadically. This fact also strongly affected the final classification of stage centres based on the frequency at which the relevant centre was included in the Tour de France itinerary as a stage place (Fig. 6)

Fig. 6

With regard to the significance we can see that many of the places included in the first editions remain continuously at the top of the rankings in terms of the quantity of performed stages (e.g., Paris, Bordeaux, Metz, Grenoble, and Lyon). However, in the recent decades, there has been a notable trend towards increasing the share of new stage places to the detriment of stable centres. This is also connected, among other things, with the above stated use of the Tour de France for the promotion of particular regions in France, which thanks to television rights can be watched by millions of viewers throughout the world and which has, naturally, a great importance in promoting tourism in the surrounding region. The stability of stage centres included in the Tour de France itinerary in selected time periods is illustrated by three chord diagrams showing the most stable stage finishes (more than 20 times in the selected time interval) whereas each diagram shows the most successful countries in terms of the number of stage triumphs (Fig. 7).

Fig. 7

The stability of stage finishes in the period before World War II (11 places with more than 20 finishes, a total of 80 stage finishes) can be seen on the diagrams. After World War II, we see an attempt of race organisers to expand the quantity of stages finishes, whereas only 3 towns (1947–1980: Bordeaux, Paris, Pau, 1981–2016: Paris, Alpe d'Huez, Pau) can be classified as stable finishes with more than 20 finishes in both following periods. However, the total quantity of stage finishes increased to 232 (1947–1980) or 338 (1981–2016), respectively. Also this graphical presentation is evidence that the organisers attempted to include the maximum possible number of stage finishes in the Tour de France itinerary. This fact is also visible in the localisation of particular finishes below the relevant diagrams on the same Figure 7. On the relevant diagrams we can also see the development of the sport dimension of the race. The clear dominance of French and Belgian cyclists in the first period is markedly balanced by the triumphs of cyclists from other countries. Based on the database with results, we can see that before 1939, cyclists from 9 countries were recorded on the scoreboard, while after 1981 it was even 32 countries

These statistics can be obtained directly from the website http://www.tdfrance.eu/custom.php.

The linking of spatial and sport dimensions is shown on the final map (Fig. 5). On this map we have only illustrated stage finishes because we covered the achievements of particular countries or of individual racers in the Tour de France stages. In the historical statistics of the Tour de France, there is a record of a total of 831 cyclists as the winners of stages (including time trials, prologues and including winning teams in team time trials). On the main map we reflected the places of the triumphs of the most successful cyclists in history in terms of the number of stage triumphs reached. The five most successful cyclists in history won a total of 139 stages, the finishes of which were in 94 different places (Fig. 8).

Fig. 8

The map above (Fig. 8) expresses the stage triumphs of individual cyclists using standard cartographic methods. The size of the diagrams represents the total number of stage finishes in observed places (in 103 editions of the Tour de France cycling race altogether). Colour filling of the diagrams shows the number of triumphs of observed cyclists in these finishes. To express the internal structure of cyclist's victories in the finish places, we used the above-described chord diagram, where the numbers of places on the map are identical to the numbers of places in the appropriate chord diagrams.

When evaluating the stage sections of particular countries, both the map and the diagram shows the victories of the five historically most successful countries (France, Belgium, Italy, the Netherlands, Spain), which reported more than 100 winning stages. The number of winning stages of these countries naturally reflects the long-term participation of cyclists from these countries in particular Tour de France editions. We must mention that in recent years, the scoreboard has been enriched by several countries in which cycling does not have a very rich history. However, globalisation also affects this sport, and therefore the movement of cyclists from various countries between professional teams with high-quality services is not exceptional. The map shows all stage finishes in the long Tour de France history. With a the coloured presentation of the number of triumphs of the countries under review, as well as with a diagram, we wanted to stress the historically most significant stage finishes, which are included in the itinerary of particular editions on a relatively regular basis. Thus we stressed 23 places, where more than 20 stage finishes were reported. In all of these stable centres, there is an apparently clear dominance of cyclists from the countries under review, whereas when looking at them in more detail, we can highlight the dominance of French and Belgian cyclists. Together, they triumphed in 1166 Tour de France stages, which is more than a 50% share of all stages performed (2215). Characteristics similar to stage finishes are also shown for the triumphs of cyclists or of particular countries in mountain ascents. It is important to stress that it is much more difficult to collect correct data for mountain ascents than for stage places. Therefore, in our analysis we used the official statistics of Tour de France, which record cyclists’ triumphs in mountain ascents of the second, first and “hors catégorie”. In this case as well we wished to highlight, in addition to a spatial aspect, the success rate of particular cyclists or of the most successful countries. For cyclists, we considered the limit to be 25 winning crossings (6 cyclists), and for the triumphs of particular countries, we considered the limit to be 200 winning crossings. (Here between Italy (the fourth) and the Netherlands (the fifth and historically most successful country), is a significant drop in the number of wins). Our analysis covers a total of 287 mountain ascents. In the presentation of the number of wins using chord diagrams, the coverage of the ascent by geomorphological unit was preferred to the total frequency of the relevant ascent included in the stage itinerary. This fact also reveals the frequency at which particular Tour de France ascents are included in the itinerary. Here we can see a frequent inclusion of stable Pyrenean ascents in comparison with Alpine hills. With respect to these particular Pyrenean or Alpine ascents which appeared first in almost the same period (1910 or 1911), the difference is mainly in the shape of the relevant geomorphological unit. The Pyrenees are line-shaped, whereas the Alpine massif is much more extensive, and thus the general variability of possible ascents is higher. The six most successful cyclists in history who are considered in our contribution triumphed a total of 205 times in 78 different ascents. A joint triumph was recorded for all cyclists in two ascents (Tourmalet and Aspin), which in the rich history of Tour de France belong to the most frequently covered ascents (Tourmalet is on the 1st place, Aspin the 3rd place). Among the wins of particular countries the situation is similar to that of stage triumphs. Historical statistics are considerably affected by the long-term participation of cyclists from traditional road cycling countries (France, Spain, Belgium, and Italy). There is also an evident marked dominance of cyclists from the countries under review in all of the observed mountain crossings (25 ascents, whereas the boundary value was 20 crossings for the analysis). An exception is the Alpe d'Huez ascent, which is, thanks to a considerable dominance of Dutch cyclists in the 1970s and in the 1980s, nicknamed “the Dutch Mountain“.

The bicycle stage race Tour de France was created in 1903. The 103 editions of Tour de France organized up till now have brought many unrepeatable sport performances which have permanently gone down in the history of sports. In addition to the sport aspect, it is also unique thanks to its spatial coverage. In addition to the first editions when particular stages repeated, we can say that each edition is unique with its spatial arrangement. The organizers attempt to gradually present new places, which with their attractiveness and visual representation during broadcasting transmission, have the potential to be the most important places for the tourism in the given region. The inclusion of 666 stage places in the Tour de France itinerary is really a unique phenomenon in the field of sporting events. This trend of extending the list of stages can be expected even in the future as this aim is also declared by the representatives of this event at the A.S.O. (Amaury Sport Organisation). In the context of stage triumphs we can also expect that the list of countries that reported a winning stage (until now 33) will grow further. Cyclists from less traditional countries are increasingly found at the start of the race. A typical example can be the participation of the first African team in Tour de France 2015 (MTN-Qhubeka).