8
Visualizing gaming trends on Steam Laurens Bolle KU Leuven [email protected] Melvin Hulsmans KU Leuven [email protected] Dries Verreydt KU Leuven [email protected] ABSTRACT Gamers and developers alike could be interested in the growth of the gaming market [10]. Will simulation games become overly abundant soon, do we see a return of arcade or retro gaming? Which genres should I develop for if I want to make a game in the most popular genre? What if I want to make one in which I have less chance of competition from other games? These questions and more can be answered by using our visualization of game genres on Steam. ACM Classification Keywords H.5.m. Information Interfaces and Presentation (e.g. HCI): Miscellaneous Author Keywords Information visualization, Gaming, Steam, Trend visualization INTRODUCTION Steam is one of the largest gaming platforms for PC [6]. It provides a platform where gamers can find, buy and play games. Even though it is one of the largest platforms already, Steam is still rapidly growing. To view and explore the con- tent of Steam in a new way, we introduce a visualization of the trends of game genres. This information visualization allows users to examine the past and current popularity of the vari- ous genres, enabling them to make informed decisions about the type of game they want to buy or develop next. GOAL AND AUDIENCE The goal of our project is to provide a visualization that helps to discover gaming trends on Steam. It shows how popular- ity of game genres evolves by visualizing the release dates of games, grouped per genre. The visualization helps game de- velopers choose what kind of games are worth making now and in the near future. A new game in a genre that has lost all its popularity probably will not be worth creating in compar- ison to a game in the genre of a booming trend. On the other hand, the visualization is also useful for developers if they want to make an original or unique game. Genres in which there have been many games released recently are probably saturated and a new game in that genre will not make that big Paste the appropriate copyright statement here. ACM now supports three different copyright statements: ACM copyright: ACM holds the copyright on the work. This is the historical ap- proach. License: The author(s) retain copyright, but ACM receives an exclusive publication license. Open Access: The author(s) wish to pay for the work to be open access. The addi- tional fee must be paid to ACM. This text field is large enough to hold the appropriate release statement assuming it is single spaced. Every submission will be assigned their own unique DOI string to be included here. of an impact. The developer is better off choosing a genre that is increasing in popularity but has not yet reached its peak. For gamers themselves, it is always fun to see in which gen- res most games are created, since they are likely to be popular and have many other players. This can help the gamer choose which game he wants to play next and as a direct consequence improve his gaming experience. TECHNOLOGY The data set Origin Initially, we were going to use the official Steam Web API [15], but unfortunately it couldn’t offer all of the infor- mation we would need for our visualization, such as release dates, total hours played or sales figures. Therefore, we use an unofficial API 1 , from which we could also only get release dates. While not being the best metric for the popularity of a game genre, it at least gives a good first indication, since we presume developers wouldn’t keep creating games for unpop- ular genres. The number of requests that can be submitted to this API in a short time period is very limited. In order to cope with this restriction, we wrote a script which periodically re- quests a small subset of the games and saves the responses in a local file, eventually resulting in a complete local data set. These requests contain a lot more information than we need, so we extracted the relevant fields and combined them into one compact JSON file. Before we were able to retrieve data from this unofficial API using this script, we did have some complications to over- come such as ‘Cross-Origin Requests’ being blocked [1]. We were able to circumvent this issue by hosting a local web server. We sent the requests from our JavaScript in a lo- cal browser to this web server, which forwarded them to the Steam server. The web server then forwarded the responses to our JavaScript where we could access this data. Format Responses from the Steam server came in a JSON format. This was easily transformed into objects in JavaScript to ex- tract the parts of the information we required, which we then saved locally again in JSON format. This made the resulting data file much smaller in size, roughly 1.5 MB. Because of the great support for JSON in D3.js, it is fast and easy to load the file whenever we start the visualization. In order to iden- tify a game, we store its Steam application ID and its name. Apart from that information, every entry has a list of genres, since it could belong to multiple genres, and its release date. 1 Documentation of the unofficial Storefront API: https://wiki. teamfortress.com/wiki/User:RJackson/StorefrontAPI

Visualizing gaming trends on Steam - student.kuleuven.bestudent.kuleuven.be/~s0218661/visualizing-gaming-trends.pdf · nal idea. Trend Visualization Genres Our next idea involved

  • Upload
    others

  • View
    6

  • Download
    0

Embed Size (px)

Citation preview

  • Visualizing gaming trends on SteamLaurens Bolle

    KU [email protected]

    Melvin HulsmansKU Leuven

    [email protected]

    Dries VerreydtKU Leuven

    [email protected]

    ABSTRACTGamers and developers alike could be interested in the growthof the gaming market [10]. Will simulation games becomeoverly abundant soon, do we see a return of arcade or retrogaming? Which genres should I develop for if I want to makea game in the most popular genre? What if I want to makeone in which I have less chance of competition from othergames? These questions and more can be answered by usingour visualization of game genres on Steam.

    ACM Classification KeywordsH.5.m. Information Interfaces and Presentation (e.g. HCI):Miscellaneous

    Author KeywordsInformation visualization, Gaming, Steam, Trendvisualization

    INTRODUCTIONSteam is one of the largest gaming platforms for PC [6]. Itprovides a platform where gamers can find, buy and playgames. Even though it is one of the largest platforms already,Steam is still rapidly growing. To view and explore the con-tent of Steam in a new way, we introduce a visualization of thetrends of game genres. This information visualization allowsusers to examine the past and current popularity of the vari-ous genres, enabling them to make informed decisions aboutthe type of game they want to buy or develop next.

    GOAL AND AUDIENCEThe goal of our project is to provide a visualization that helpsto discover gaming trends on Steam. It shows how popular-ity of game genres evolves by visualizing the release dates ofgames, grouped per genre. The visualization helps game de-velopers choose what kind of games are worth making nowand in the near future. A new game in a genre that has lost allits popularity probably will not be worth creating in compar-ison to a game in the genre of a booming trend. On the otherhand, the visualization is also useful for developers if theywant to make an original or unique game. Genres in whichthere have been many games released recently are probablysaturated and a new game in that genre will not make that big

    Paste the appropriate copyright statement here. ACM now supports three differentcopyright statements:• ACM copyright: ACM holds the copyright on the work. This is the historical ap-proach.• License: The author(s) retain copyright, but ACM receives an exclusive publicationlicense.• Open Access: The author(s) wish to pay for the work to be open access. The addi-tional fee must be paid to ACM.This text field is large enough to hold the appropriate release statement assuming it issingle spaced.Every submission will be assigned their own unique DOI string to be included here.

    of an impact. The developer is better off choosing a genre thatis increasing in popularity but has not yet reached its peak.For gamers themselves, it is always fun to see in which gen-res most games are created, since they are likely to be popularand have many other players. This can help the gamer choosewhich game he wants to play next and as a direct consequenceimprove his gaming experience.

    TECHNOLOGY

    The data setOriginInitially, we were going to use the official Steam WebAPI [15], but unfortunately it couldn’t offer all of the infor-mation we would need for our visualization, such as releasedates, total hours played or sales figures. Therefore, we usean unofficial API1, from which we could also only get releasedates. While not being the best metric for the popularity of agame genre, it at least gives a good first indication, since wepresume developers wouldn’t keep creating games for unpop-ular genres. The number of requests that can be submitted tothis API in a short time period is very limited. In order to copewith this restriction, we wrote a script which periodically re-quests a small subset of the games and saves the responses ina local file, eventually resulting in a complete local data set.These requests contain a lot more information than we need,so we extracted the relevant fields and combined them intoone compact JSON file.

    Before we were able to retrieve data from this unofficial APIusing this script, we did have some complications to over-come such as ‘Cross-Origin Requests’ being blocked [1]. Wewere able to circumvent this issue by hosting a local webserver. We sent the requests from our JavaScript in a lo-cal browser to this web server, which forwarded them to theSteam server. The web server then forwarded the responsesto our JavaScript where we could access this data.

    FormatResponses from the Steam server came in a JSON format.This was easily transformed into objects in JavaScript to ex-tract the parts of the information we required, which we thensaved locally again in JSON format. This made the resultingdata file much smaller in size, roughly 1.5 MB. Because ofthe great support for JSON in D3.js, it is fast and easy to loadthe file whenever we start the visualization. In order to iden-tify a game, we store its Steam application ID and its name.Apart from that information, every entry has a list of genres,since it could belong to multiple genres, and its release date.1Documentation of the unofficial Storefront API: https://wiki.teamfortress.com/wiki/User:RJackson/StorefrontAPI

    https://wiki.teamfortress.com/wiki/User:RJackson/StorefrontAPIhttps://wiki.teamfortress.com/wiki/User:RJackson/StorefrontAPI

  • Distribution of valuesThe data retrieved dated back from 1997 to now, as shown inFigure 1. Only 26 games dated back before 2005, this is whywe removed all the data leading up to the year 2005.

    Figure 1: Visualization including all the years that containedany data.

    Since 2012 Steam also accepts non-gaming applications [17].This led to our data containing nine genres that are not gam-ing genres but rather tags for the type of software. Examplesof these genres are Web Publishing and Photo Editing. Anold version of the visualization with all the genres is shown inFigure 2. The removed genres contained around two percentof all our data points. Removing them made the visualizationmore clear without losing valuable information.

    Figure 2: Visualization including all the genres found in thedata.

    The eight most common genres have a very similar distribu-tion from 2005 to 2015. Slowly increasing over the earlieryears with a larger increase the closer we get to the present.All genres have a very large increase in 2014 and they allhave their current maximum in 2014. The data for 2015 isonly partially collected, but for most genres it already is overhalf the value of 2014. The less common genres have zerovalues for the earlier years, but there was enough data to startfrom 2005.

    When the data was grouped up in quarters of a year, the highervalues within a year were almost always the fourth quarterwith a lot lower values for the first and second quarter of theyear.

    The less common genres show more irregularities from thesedistributions because there is less data to work with.

    D3.jsSince this was our first time programming in D3, we took alook at the examples page of D32. Here we found a couple oftechniques that we decided to use for ourselves. It should benoted that some code is copied and edited to work with ourvisualization, as detailed below.

    We used the example on http://bl.ocks.org/mbostock/3887051 as a base for our grouped bar chart. The codeon http://bl.ocks.org/mbostock/3943967 taught us howto work with radiobuttons, transitions and how to make astacked bar chart. The method to construct our y-axis lineswas inspired by http://bl.ocks.org/mbostock/4323929.After combining all of these components we still needed toimplement a line graph, for which we took the following twosites as examples: http://bl.ocks.org/mbostock/3883245,http://bl.ocks.org/mbostock/3884955. Connecting theseexample visualizations, customizing them to our liking andmaking them compatible with our data format proved to bea considerable challenge. For example, our data consisted ofa long list of games with their respective attributes. But thestacked bar chart required the data to be organized in a listwith each element representing a year containing the numberof games per genre.

    VISUALIZATION AND INTERACTION

    Road to the visualization

    Achievement visualizationInitially we wanted to focus on ‘achievements’. An achieve-ment of a game is a badge that you get on your Steam ac-count showing that you have achieved a certain feat. Thesefeats can range from completing the tutorial to performing anearly impossible action. We wanted to visualize how manyachievements of a certain game were actually achieved by theentire community. This could enable so-called ‘achievementhunters’ to find the more ‘valuable’ achievements, since theywould not have been completed by many other users. We alsowanted to add a login function that would work with the steamaccount. This would reduce the visualization to only show thegames that the user owns. The way we wanted to visualizethis was by showing a cluster of orbs where each orb wouldcontain the cover art and name of a certain game. When a userwould click on a certain game’s orb, all the other orbs woulddisappear and the achievements of this specific game wouldsurround the orb. Every achievement’s icon would be in aseparate orb and would contain the percentage and amount ofplayers that obtained it.

    However, Steam already provides many of these features in-side the client itself. While Steam doesn’t provide the infor-mation in an interactive visualization, recreating these con-cepts seemed redundant unless we could add a significantnumber of novel features. We also wanted the visualization to

    2The D3.js examples gallery can be found at: https://github.com/mbostock/d3/wiki/Gallery

    http://bl.ocks.org/mbostock/3887051http://bl.ocks.org/mbostock/3887051http://bl.ocks.org/mbostock/3943967http://bl.ocks.org/mbostock/4323929http://bl.ocks.org/mbostock/3883245http://bl.ocks.org/mbostock/3884955https://github.com/mbostock/d3/wiki/Galleryhttps://github.com/mbostock/d3/wiki/Gallery

  • allow users to explore the data and gain new insights, some-thing which would be difficult when only displaying achieve-ments. Due to these problems, we decided to drop our origi-nal idea.

    Trend Visualization GenresOur next idea involved visualizing the evolution of the var-ious game genres available in the Steam store. This subjectallows for users to explore the data and possibly find storiesor interesting questions such as: “When did Indie games be-come popular?”

    We started by looking for methods to visualize this data. Thefirst visualization that we found was the stacked area chart.Figure 13 shows an example of this kind of chart. This vi-sualization is good to portray a trend throughout the years ifyou have one or more data sets. Our sets are the genres thathad a certain number of games released in a certain year. Un-fortunately stacked area charts have some problems as well,as explained in the ‘Related work’ section, which is why wedidn’t use this technique.

    Figure 3 is another visualization we considered, based on thework of Rahman [11]. This visualization makes it easy to ei-ther compare the different genres in a single year, or view theevolution of a single genre throughout the years. The problemhowever is the growth of the Steam platform, which is verydifficult to view using this visualization. For example, the to-tal number of games more than doubled in 2014 compared to2013. Deducing this information would be difficult for a userwhen using this visualization.

    Figure 3: The number of released games per year, separatedby genre, visualized using the technique of Rahman [11].

    Figure 4 shows a D3.js recreation [4] of the famous Gap-minder [12] visualization. Where the original dynamicallyplots wealth versus life expectancy over time in which theorbs represent countries, we could plot number of releasesversus number of players with the orbs representing the dif-ferent genres. We opted against using this visualization sinceSteam only exists since 2003 [14], so the animation would

    be somewhat limited. Furthermore, no data is publicly avail-able about the number of players, nor could we find any otherinteresting, and available, metrics to plot.

    Figure 4: D3.js recreation [4] of the Gapminder [12] visual-ization.

    Finally we also considered using a stacked area chart, as il-lustrated in Figure 13. The ‘Related works’ section discussessome problems with this visualization, such as the difficultyof viewing the evolution of one of the higher layers. Due tothe problems with the stacked area chart, we decided to baseour visualization on bar charts.

    Different ViewsOur visualization has different views. We wanted to be able todisplay all the information that can be obtained from the data.Trying to make one view that could do this turned out to makethe visualization unclear. This is also a bad working methodaccording to Few [7]. This could mean that the user isn’t ableto see certain information or stories that could be obtainedfrom the data. Just choosing the most important view is alsonot a solution. This way stories might, and in our case will belost. A good example of this is when a story can be found bylooking at multiple views.

    Grouped bar chartThe first view we will discuss is the grouped bar chart. Ourfirst implementation of this can be found in Figure 5. Thisvisualization enables the user to compare genres within oneyear. You also still retain a general overview of all the genresand their growth throughout the years. However the overalltrend of a genre throughout the years is rather difficult to see.

    Our final visualization can be found in Figure 6. We can seehere that the legend is expanded with checkboxes. These al-low the user to choose the genres that are displayed. Selectingor deselecting genres results in rescaling. By doing this thereis less loss of screen space. The rescaling also allows the userto closer inspect smaller genres. When deselecting large gen-res the smaller genres will be displayed a lot larger as can beseen in Figure 7.

    We also added numbers on top of the bars. These will alsorescale. They were added because the exact amount of thegames with a certain genre that were released on a year couldnot be determined from the y-axis. Of course this was only

  • Figure 5: First implementation

    Figure 6: Grouped bar charts showing the absolute number ofgames for each genre per year

    (a) Large genres selected

    (b) Large genres deselected

    Figure 7: Visual example of rescaling

    (a) Mockup

    (b) Final visualization

    Figure 8: Stacked bar charts showing the absolute number ofgames for each genre

    the case if the amount of games was sufficiently high enough.The intended use of the y-axis is to be able to get a roughindication of the amount.

    Stacked bar chart with absolute numbersThe absolute view was made to show the growth of Steam andits genres. Before implementing this we first made a mockup,which can be found in Figure 8a. Our final visualization canbe found in Figure 8b. It can be seen that again the check-boxes are added to the legend. In the mockup they were sepa-rate but we combined them to increase the data-ink ratio. Wechose to put the color of the genre behind its name. This be-cause when merging the legend and the checkboxes the colorswould fall behind the checkboxes. Selecting and deselectinggenres will rescale the genres to use as much screen space aspossible.

    Another difference is that we removed the search bar, sincethere is enough space to display all the genres and still keepthem readable. We also added the exact amount of genresinside of the bars. The numbers will scale relative to the totalamount of games. These numbers were also added becausethe exact amount, of the games of a certain genre that werereleased in a year, could not be determined from the y-axis.This again only applies if the amount of games is sufficientlyhigh.

    Another thing we added is an on-click event. This can beseen in Figure 9. We added this because like in a stacked areachart, viewing the evolution of one of the higher layers in astacked bar chart is difficult, this due to the different baselines

  • (a) Action on the bottom of the graph

    (b) Strategy on the bottom of the graph

    Figure 9: Visual example of the move-to-bottom on-clickevent

    caused by the bottom layer. The click-event will move thebars of the genre that was clicked on to the bottom. This wasour way to remove one of the biggest problems with a stackedbar chart.

    Stacked bar chart with relative numbersBecause the platform of steam has been growing a lot in thelast couple of years the user needs a view that allows him toclearly see how the share of a genre has grown or shrunk. Toget this information from the absolute or relative view wouldalready be difficult if the total amount of games stayed thesame each year. However with the constant growth of steamand the birth of new genres it is hard to just see this. Becauseof this we added a view which made it immediately clear howa genre has grown in comparison to the other genres.

    Our first mockup can be found in Figure 10a and the final vi-sualization can be seen in Figure 10b. Here again we can seethat the legend has been changed, checkboxes and numbershave been added and the search bar has been removed. Thereasons and how the rescaling works are the same as in theStacked bar chart with absolute numbers section. Again theon-click sorting of the bar charts was added to counter theproblems with the stacked charts.

    The data as well as the numbers are now displayed in per-centages. When deselecting genres the non-selected genreswill be displayed in gray. This to show that selecting all thegenres will result in displaying 100% of the games.

    (a) Mockup

    (b) Final visualization

    Figure 10: Stacked bar charts showing the number of gamesrelative to the total number of games per year

    Comparing genres in detailThe last view of our visualization is intended for the user tocompare different genres with each other in more detail. Todo this we increased the granularity of the time axis to quar-ters of a year and display the data using lines instead of bars.This can be seen in the mock-up in Figure 11a and the finalvisualization in Figure 11b. Again the checkboxes and leg-end were added and merged. The search bar was again foundredundant. The rescaling also works in this view.

    We displayed the lines of the unchecked genres in gray. Onereason we do this is to give the user an indication about wherethe currently selected lines are positioned in the general data.The main reason was to indicate that there is still data to bedisplayed. We did not do this in the absolute bar charts andgrouped bar charts because the rescaling does not allow this.The reason for this is that bars of the grouped bar chart wouldoverlap with the grayed data and the absolute bar chart’s barwould push it off the screen. In the line charts however therescaling will also move the lines off the screen but the slopewill indicate the general trend it will go through. If the userwants more information he can just select the genre. Becausea lot of lines are close to each other it is not a good idea toselect too many genres at the same time. It is better to selecta small amount and make use of the automatic rescaling tocompare even the small genres.

    InteractionAs shown and explained earlier the user can interact withthe bars or lines by selecting or deselecting genres, choos-

  • (a) Mockup

    (b) Final visualization

    Figure 11: Line graph showing the evolution of genres perquarter, instead of per year

    ing views through the radio buttons and clicking on the barsin the stacked bar charts.

    An interaction that is not explained yet is hovering over thelegend, bars or lines, which is integrated in all the graphs.When a user hovers over a certain genre in the legend its cor-responding bars are given a frame as in Figure 12a in case wehave bar chars. In case of the line chart the lines are high-lighted as seen in Figure 12b, here the purple line is the high-lighted one.

    When the user hovers over a bar or line the genre correspond-ing to the data that is hovered over will be given a frame. Inthe bar chart this will look the same as Figure 12a with theexception of the location of the mouse cursor. In the compareview this will look the same as Figure 12b with the exceptionof the mouse cursor. Also the line that is hovered over willnot turn purple.

    RELATED WORKAlthough Steam was initially released in 2003 [14] and iscurrently one of the largest PC gaming platforms [6], aca-demic research of the various aspects of this community hasbeen scarce. Blackburn et al. [3] analyzed the behaviourof cheaters in the Steam gaming community. Meanwhile,Becker et al. [2] claim to be the first to analyze Steam’s so-cial network in 2012, showing it behaves very similar to otherpopular social networks. Besides these examples, Steam has,to the best of our knowledge, seldom been the focus of aca-demic research, especially in the field of information visual-ization and has thus been of little influence on our work.

    (a) Hover over legend in relative

    (b) Hover over legend in line graph

    Figure 12: Example visualizing the hover over interaction

    Outside of academic research, there are some visualizationson the web targeting Steam. An interesting example, that isvery related to our own work, is Steam Charts [13]. Thisweb page has been tracking the number of concurrent play-ers on an hourly interval for every game in the Steam catalogsince July 2012. With interactive line graphs and numbers,it provides a very good overview of the popularity of variousgames. However it is not possible to compare different gamesor view the evolution of game categories or genres. Our visu-alization specifically targets the evolution of the popularity ofdifferent game genres and provides various views to comparethem.

    Unrelated to Steam or gaming in general, Wattenberg [16]created a very popular web-based visualization applet, the, which enables users to interactively explore name data -specifically, the evolution of the popularity of specific namesover time. Although the topic is very different, the goal of thevisualization is very similar to ours: allowing users to exploreand discover certain trends in the data. As illustrated in Fig-ure 13, NameVoyager uses a stacked area chart to visualizeits numbers.

    We chose to use stacked bars, instead of a stacked area chartfor our visualization, because people are generally bad atestimating areas [8]. Müller and Schumann [9] also men-tion that stacked bar charts are good representations for time-dependent data and according to Mackinlay [8], length is amore accurate representation of quantitative data, which fur-

  • Figure 13: NameVoyager visualization of baby names start-ing with ‘Jo’. Using a stacked area chart the popularity ofthe different names since the 1900s is shown. (Source im-age: [16])

    ther strengthens our decision for using stacked bar chartsrather than area based representations. The NameVoyager al-lows for hovering over a layer to highlight it and clicking alayer removes the other layers, giving you a clear overviewof the evolution of that specific name. This highlighting andselecting is a very useful feature when dealing with stackeddata, since a bad ordering of the data makes it hard to view theevolution of one of the higher layers [5]. Therefore, we im-plemented a very similar feature where we, instead of remov-ing all of the other layers, reorder them and put the selectedone as the bottom one, similar to the baseline reordering assuggested by Yi et al. [18]. While we have chosen to use a lotof stacked bar charts, Few [7] argues that line graphs are evenbetter for showing multiple time-series, which is why we alsoprovide a line graph with more detailed information.

    In Yi et al. [19], four pillars for making people gain in-sights using information visualization are provided: 1) Pro-vide Overview, 2) Adjust, 3) Detect Pattern, and 4) MatchMental Model. We aim to provide insights for gamers anddevelopers into the different game genres on Steam with ourvisualization. Therefore we made sure to incorporate thesefour pillars. We provide overview (1) by displaying the evo-lution of the complete data set in various charts. The user canadjust these charts (2) by reordering the layers, (de)selectingadditional game genres or by changing the view. It is then upto the user to detect patterns in the data (3). The visualizationsupports the user in this step by providing a visual represen-tation of the data, which decreases the gap between the dataand user’s mental model (4) of it, thereby reducing cognitiveload in understanding [19].

    OUTRO: WHAT COULD BE DONE BETTER?Because we did not start from an existing data set we useda lot of our programming and research time to find, createand use the data we needed. We used the release dates ofgames as a metric because this was the only metric we couldfind and retrieve. We would have liked to use better metricssuch as sales numbers or amount of hours played for eachgame to better estimate popularity of games and their genres,but unfortunately they were not publicly available. Perhaps

    after discovering this, we should have considered completelychanging subjects and look for a better data set.

    A minor thing we could have changed was made clear to us inthe end, the colors of the numbers on the visualization couldbe made dynamic to contrast better to the colors of the visu-alization elements.

    Tool tips could be added to cover information that is notshown because the size of the visualization elements is toosmall. All the information could be present without beingunreadable, just hidden behind some interaction to keep theoverview clean.

    CONCLUSION & FUTURE WORKWe made a visualization of the Steam platform which, to thebest of our knowledge, did not exist yet. Current developersand Steam users might be interested in our result. We believethat with a better data set and less problems along the road,the visualization could have reached its full potential. Thisof course doesn’t take away that the current visualization canbe used to find interesting facts about the Steam platform andteach the users more about the trends of genres.

    The experience was very educational both on the language ofD3.js, which required a different mindset to program in andthe lessons about visualizations that showed us which ideaswere good and more importantly which ones were not.

    Future workA feature we wanted to implement but decided not to becauseof time constraints is a search function for a specific game. Itwould highlight the specific genre and year the data point ofthe game would be in.

    Another more personalized feature we considered was allow-ing the user to log in with his Steam account to place theplayer into the global statistics. Making the user able to com-pare himself to others and situate himself on the Steam plat-form. Because of the way we used the Steam API howeverwe were not able to implement this.

    Finally, as mentioned in the Outro, adding a tooltip when hov-ering over the line graphs to show the exact data points, couldhave been a valuable addition. Due to time constraints thisfeature was also dropped.

    REFERENCES1. Adam Barth. 2011. The web origin concept. (2011).

    2. Roi Becker, Yifat Chernihov, Yuval Shavitt, and NoaZilberman. 2012. An analysis of the steam communitynetwork evolution. In Electrical & ElectronicsEngineers in Israel (IEEEI), 2012 IEEE 27thConvention of. IEEE, 1–5.

    3. Jeremy Blackburn, Ramanuja Simha, Nicolas Kourtellis,Xiang Zuo, Clayton Long, Matei Ripeanu, JohnSkvoretz, and Adriana Iamnitchi. 2011. Cheaters in theSteam Community Gaming Social Network. ArXive-prints dec (2011). http://adsabs.harvard.edu/abs/2011arXiv1112.4915B$\delimiter"026E30F$nhttp://arxiv.org/abs/1112.4915

    http://adsabs.harvard.edu/abs/2011arXiv1112.4915B$\delimiter "026E30F $nhttp://arxiv.org/abs/1112.4915http://adsabs.harvard.edu/abs/2011arXiv1112.4915B$\delimiter "026E30F $nhttp://arxiv.org/abs/1112.4915http://adsabs.harvard.edu/abs/2011arXiv1112.4915B$\delimiter "026E30F $nhttp://arxiv.org/abs/1112.4915

  • 4. Mike Bostock. 2012. The Wealth & Health of Nations.(2012). Retrieved May 12, 2015 from ,http://bost.ocks.org/mike/nations/.

    5. Lee Byron and Martin Wattenberg. 2008. Stackedgraphs - Geometry & aesthetics. IEEE Transactions onVisualization and Computer Graphics 14, 6 (2008),1245–1252. DOI:http://dx.doi.org/10.1109/TVCG.2008.166

    6. Oliver Chiang. 2011. The Master of Online Mayhem.(2011). Retrieved May 12, 2015 fromhttp://www.forbes.com/forbes/2011/0228/technology-gabe-newell-videogames-valve-online-mayhem.html.

    7. Stephen Few. 2011. Quantitative Display for CombiningTime-Series and Part-to-Whole Relationships. (January,February, and March 2011).

    8. Jock Mackinlay. 1986. Automating the design ofgraphical presentations of relational information. ACMTransactions on Graphics 5, 2 (1986), 110–141. DOI:http://dx.doi.org/10.1145/22949.22950

    9. W. Muller and H. Schumann. 2003. Visualizationmethods for time-dependent data - an overview.Proceedings of the 2003 Winter Simulation Conference,2003. 1 (2003). DOI:http://dx.doi.org/10.1109/WSC.2003.1261490

    10. Newzoo BV. 2014. Global Games Market Will Reach$102.9 Billion in 2017. (2014). Retrieved May 12, 2015from , http://www.newzoo.com/insights/global-games-market-will-reach-102-9-billion-2017-2/.

    11. Asif Rahman. 2013. Journals. (2013). Retrieved May 12,2015 from ,http://neuralengr.com/asifr/journals/.

    12. Hans Rosling. 2009. Gapminder. GapMinderFoundation http://www.gapminder.org (2009), 91.

    13. Steam Charts. 2012. An ongoing analysis of Steam’splayer numbers. (2012). Retrieved May 11, 2015 fromhttp://steamcharts.com/.

    14. Valve Corporation. 2013. Steam Client Released.(2013). Retrieved April 25, 2015 fromhttp://store.steampowered.com/news/183/.

    15. Valve Corporation. 2015. Steam Web APIDocumentation. (2015). Retrieved April 25, 2015 fromhttp://steamcommunity.com/dev.

    16. Martin Wattenberg. 2005. Baby names, visualization,and social data analysis. In Information Visualization,2005. INFOVIS 2005. IEEE Symposium on. IEEE, 1–7.

    17. Wikipedia. (????).

    18. Ji Soo Yi, Youn Ah Kang, John Stasko, and Julie Jacko.2007. Toward a deeper understanding of the role ofinteraction in information visualization. IEEEtransactions on visualization and computer graphics 13,6 (2007), 1224–31. DOI:http://dx.doi.org/10.1109/TVCG.2007.70515

    19. Ji Soo Yi, Youn-ah Kang, John T Stasko, and Julie aJacko. 2008. Understanding and characterizing insights:How Do People Gain Insights Using InformationVisualization?. In Proceedings of the 2008 conferenceon BEyond time and errors novel evaLuation methodsfor Information Visualization - BELIV ’08. 1. DOI:http://dx.doi.org/10.1145/1377966.1377971

    http://bost.ocks.org/mike/nations/http://dx.doi.org/10.1109/TVCG.2008.166http://www.forbes.com/forbes/2011/0228/technology-gabe-newell-videogames-valve-online-mayhem.htmlhttp://www.forbes.com/forbes/2011/0228/technology-gabe-newell-videogames-valve-online-mayhem.htmlhttp://www.forbes.com/forbes/2011/0228/technology-gabe-newell-videogames-valve-online-mayhem.htmlhttp://dx.doi.org/10.1145/22949.22950http://dx.doi.org/10.1109/WSC.2003.1261490http://www.newzoo.com/insights/global-games-market-will-reach-102-9-billion-2017-2/http://www.newzoo.com/insights/global-games-market-will-reach-102-9-billion-2017-2/http://neuralengr.com/asifr/journals/http://steamcharts.com/http://store.steampowered.com/news/183/http://steamcommunity.com/devhttp://dx.doi.org/10.1109/TVCG.2007.70515http://dx.doi.org/10.1145/1377966.1377971

    IntroductionGoal and audienceTechnologyThe data setOriginFormatDistribution of values

    D3.js

    Visualization and interactionRoad to the visualizationAchievement visualizationTrend Visualization Genres

    Different ViewsGrouped bar chartStacked bar chart with absolute numbersStacked bar chart with relative numbersComparing genres in detailInteraction

    Related workOutro: what could be done better?Conclusion & Future workFuture work

    REFERENCES