Assignment 1 - Documentation
NRL 1998-2008: Comparison between End of Year Rankings and Crowd Sizes of 13 Teams.
Data Characteristics
Looking at each team, different trends start to appear. Some are quite unique, but others appear quite consistently between different teams. These data sets compare end of year rankings to overall home crowd sizes for that particular year between 1998 and 2008. The main reason for this is an assumption I have, that if a team performs consistently well, crowds will consistently attend games.
Brisbane Broncos
Season Ranking Total Crowds
1998 1 240878
1999 8 278791
2000 1 276082
2001 5 256780
2002 3 241575
2003 8 289023
2004 3 344007
2005 3 363923
2006 3 374494
2007 8 394415
2008 5 401108
Starting with the Brisbane Broncos, it can be seen from 1999, the crowd size begins to decrease and from 2003 onwards, the crowd size continuously increases. The same cannot be said about their rank, this does not follow in the same trend; even at the points where the team was ranked 1st, crowd numbers during that season where nowhere near the highest points. The team seems to have maintained a stable ranking in this 11 year period and the do not drift past the number 8 spot. In the sport of rugby league, teams strive to reach a top 8 position as this is the only way to move through to the finals (other teams being eliminated instantly). I may be said that as a result of this consistent performance, fans have stood by their team and crowd numbers are growing to some of the highest numbers ever recorded.
St George Illawarra Dragons
Season Ranking Total Crowds
1998 8 124728
1999 6 263692
2000 9 132884
2001 7 153772
2002 7 149389
2003 10 156911
2004 5 184058
2005 2 177358
2006 6 183182
2007 13 145908
2008 7 151154
The Dragons are one team which seam to more or less follow this assumption that team performance is influenced by crowd size. Nearly every instance where ranking improves, crowd numbers move forward with it. There is one instance, in 1999, where crowd number where at the highest point in 11 years. This suggests that something may have occurred during this year to bring about such high crowd numbers. In actual fact, this was the year the Dragons made it to the grand final match which may have had great influence in this data.
Newcastle Knights
Season Ranking Total Crowds
1998 2 243518
1999 7 250563
2000 3 243473
2001 3 228275
2002 2 222224
2003 7 217374
2004 9 207105
2005 15 221628
2006 4 262179
2007 15 190563
2008 9 224996
The Knights are a team which have both fluctuating crowd numbers and ranking and both of these data sets are inconsistent with each other. Like the Broncos, the Knights are the kind of team which maintain a steady rank. However, when they begin to drift outside the top 8, this is where we start to see the lower end of their crowd number and because of this assumption of rank affecting crowd size stands.
New Zealand Warriors
Season Ranking Total Crowds
1998 15 120478
1999 11 135265
2000 13 161042
2001 8 165940
2002 1 189082
2003 6 201536
2004 14 122025
2005 11 157212
2006 10 105946
2007 4 158699
2008 8 137325
This is the team which by far shows the greatest favourability for the assumption of rank affecting crowd size. Working down from one year to the next it can be seen that when rank improves, crowd numbers increase accept for 2003 where they reach and 11 year high. This may be due to their great performance in the pervious year where they came first, carrying on good performance in the following year.
North Queensland Cowboys
Season Ranking Total Crowds
1998 16 204981
1999 16 193296
2000 14 201771
2001 13 170732
2002 11 145600
2003 11 178709
2004 7 200357
2005 5 247075
2006 9 236175
2007 3 241262
2008 15 217228
This is a team which showed poor performance up until 2004 and after this point, not only does the rank increase, but the crowd numbers move up as well. Even in 2008 where they placed poorly, the Cowboys maintained a steady crowd size, demonstrating, like the Broncos, that crowd numbers are increasing.
Sydney Roosters
Season Ranking Total Crowds
1998 6 123104
1999 4 156380
2000 2 201426
2001 6 162961
2002 4 137389
2003 2 193161
2004 1 212053
2005 9 192681
2006 14 154151
2007 10 143421
2008 4 167642
Like the Warriors, the Roosters show great consistency between the two data sets. End of year rankings seems to definitely affect the crowd sizes for this club. This is supported by the fact that the points where the crowd numbers were at their highest, above 200 000, are when the team was place 1st and 2nd.
Manly Sea Eagles
Season Ranking Total Crowds
1998 10 126155
1999 13 215521
2000 12 169963
2001 10 133111
2002 9 112093
2003 14 124404
2004 13 134435
2005 8 182054
2006 5 189522
2007 2 181084
2008 2 164324
The assumption of ranking effecting crowd size seems to have a negative effect in the beginning of this data set. Their crowd numbers do seem to rise in the later part of the data but as their rank increases to an 11 year high point, crowd numbers begin to slowly fall. It could be said that the team does have support, but these fans do not like to go to home games. Only when the team’s performance starts to slip, the fans turn out to help give their team a push.
Penrith Panthers
Season Ranking Total Crowds
1998 14 111092
1999 10 149939
2000 5 186771
2001 14 131757
2002 12 132137
2003 1 212419
2004 4 211049
2005 10 186906
2006 12 138953
2007 16 144419
2008 12 130790
The Panther’s data shows that most of the time, the two data sets are consistent with each other. What’s more is that once again high ranking has draw the highest crowd numbers and both the performance and crowd numbers were maintained in the following year.
Cronulla Sharks
Season Ranking Total Crowds
1998 11 136444
1999 1 168295
2000 8 238974
2001 4 161513
2002 5 160890
2003 12 154738
2004 11 149865
2005 7 195371
2006 13 149859
2007 11 128302
2008 3 155581
The Sharks do show some consistency between the two datasets of rank and crowd numbers. What’s more interesting is fact the once again great performance in one year has lead to great crowd numbers in the following year. In 1999 the Sharks came 1st and in the following year, their crowd numbers hit an 11 year high with 238974. From this, it can be said after the team had a terrific season, the fans wanted to be there to try and help replicate this performance by supporting their team.
Canberra Raiders
Season Ranking Total Crowds
1998 7 118303
1999 9 144744
2000 4 158054
2001 11 124469
2002 8 125533
2003 4 160772
2004 8 133391
2005 14 148851
2006 7 137870
2007 14 138144
2008 6 142952
The Raiders don’t show much consistency between the two data sets. However like the Roosters and the Panthers, the greatest crowd numbers appear where rank was highest. So although there wasn’t any real consistency form one year to the next, higher ranking still produces greater rank maintaining come consistency in data correlation assumption.
Melbourne Storm
Season Ranking Total Crowds
1998 3 152688
1999 3 154827
2000 6 190092
2001 9 155745
2002 10 108693
2003 5 115509
2004 6 106583
2005 6 106770
2006 1 130426
2007 1 140527
2008 1 149686
Although crowd numbers are quite low for this team, the two data sets are still quite consistent with each other. The only place where this isn’t true is 1999, where rank decreases but crowd numbers hit an 11 year high. This does suggest that something may have occurred during this year to draw more crowds. In actual fact, 1999 was the year where Melbourne won the Grand Final, and this most likes would have been the cause for such high crowd numbers; fans wanting to witness, first hand, a repeat of results.
Parramatta Eels
Season Ranking Total Crowds
1998 4 136310
1999 2 174604
2000 7 186190
2001 1 271600
2002 6 169053
2003 9 135142
2004 12 141186
2005 1 195169
2006 8 175042
2007 5 169706
2008 11 162335
The Eels show some consistency between the two data sets, but once again, the point where crowd numbers where greatest was where rank was highest. This brings about the assumption that if a team is winning, the fans will come to games to support their team.
Canterbury Bulldogs
Season Ranking Total Crowds
1998 9 100358
1999 5 203450
2000 11 167489
2001 2 187557
2002 15 169919
2003 3 254280
2004 2 216382
2005 12 219431
2006 2 216489
2007 6 197852
2008 16 156564
The Bulldogs show a fair amount of consistency between the two data sets. Most of the lower ranks seem to generate the lowest crowd numbers. From 2006, you see a gradual slip of both crowd size and end of year ranking.
Looking at all these datasets it can be seen that in the majority of cases, end of year rankings will determine crowd size. Some teams demonstrate this quite consistently one a year to year basis. Others maintain this assumption due to the fact that when the team performs better then most in the comp (i.e. top 3 ranking), crowd number peak for these teams.
It can also be seen that team who do not reach this top rank do not achieve these high crowd numbers regularly. It is as if once it occurs, fan support rises, boosting the team to try repeat great performance. For example the Dragons, Raiders and Eagles are 3 teams without a number 1 spot in the past 11 years and their crowd numbers are lower then most other teams.
Comparing in the same region, it can be seen that although the Brisbane Broncos and North Queensland Cowboys have quite different rankings and crowd numbers, the distribution and progression of the crowd size is quite similar. The crowd numbers for both teams start off quite steady increasing or decreasing by small increments, and then suddenly, in 2004; crowd numbers for both teams start to dramatically increase. These are the only tow teams with this same structure which would suggest a great increase in fan support in the state over the past few years.
The Melbourne Storm are a team which stand out from the others. Unlike any other team, they have consistently high ranking but some of the lowest crowd numbers out of any other team. For a team which has scored more 1st rankings in the past 11 year than any other team this does suggest that something is wrong. Fact of matter is that in Victoria, AFL is the sport of choice, dominating the state, leaving room for a lot less supports of rugby league. The NZ Warriors is another team which demonstrates this. Outside influence of rugby union once again drains the fan base for rugby league.
Low crowd numbers could be caused by something else. The teams with the lowest crowd numbers include the Canberra Raiders, The Melbourne Storm and the NZ Warriors. These teams have no competition close by (geographically) which brings about further assumptions. Supporters of these teams may not be as fierce as others in NSW and OLD and many of them don’t have to constantly deal with fans that support other teams. It could also be that when these teams have how games, fans of the opposing team are not willing to make an extended trip to a game interstate, or in another country.
Design Rationale
The concept for the data model began with 2 datasets; end of year ranking and crowd numbers for all the teams in the NRL from 1998 to 2008. The reason for the 11 year period is the fact that before 1998 there were complications in the league; some teams were folding others was merging. Another reason for this short time period is to take a more modern look at the game. Looking at a sport now compared to 20 years ago, no matter what the sport is, the sport would have always changed as a result of greater support, funding, and human abilities.
Running through and comparing these two datasets against time allowed for one team to stand out, the NZ Warriors; where a change in one dataset was consistent with the change in the other, moving though one year to the next.
The hardest thing was now designing a data-mapping technique which allowed each dataset to be dependent on one another (i.e. if one of the two datasets was removed, the shape and form of the remaining data set will change).
Starting with simples shapes, a sculpture was created. Using cylinders to represent each year, each cylinder was stacked on top of the other representing an annual increase with each new cylinder. The net 2 attributes affected the shape of each cylinder. Crowd size determines the radius. The larger the crowd size, the greater the radius. End of year rank determined the height/thickness of each cylinder. The greater the rank, the thicker the cylinder. A scaling system was created so that in Maya, cylindrical thickness values could easily and accurately be changed. The scale is as follows:
1st – 1.6
2nd – 1.5
3rd – 1.4
4th – 1.3
5th – 1.2
6th – 1.1
7th – 1
8th – 0.9
9th – 0.8
10th – 0.7
11th – 0.6
12th – 0.5
13th – 0.4
14th – 0.3
15th – 0.2
16th – 0.1
In Maya it became hard to determine between cylinders (i.e. one year from the next) and so spacers were created between each cylinder.
After looking at the sculpture that was created, I found the dataset to be quite small, and not much meaning could be formed from a single team. Comparisons were needed between other teams to allow for greater reliability (i.e. finding similarities between different teams in order to support a given meaning). The reason for starting with only one team’s datasets was because it was the one which showed the most consistency between the two datasets from one year to the next. Now the question that was why are they the most consistent team? Sculptures were then created for every other team in the competition who has competed in every season for the past 11 years. Other teams were left out because they could cause irregularities in the final sculpture in all the team-sculptures were combined.
The order of these sculptures from left to right is bulldogs, eels, storm, raiders, sharks, panthers, eagles, roosters, cowboys, warriors, knights, dragons, broncos.
Each sculpture seemed to have its own unique shape and told its own story, but side by side, other patterns started to appear (discussed above).
Now the challenge became the integration of all 13 models into one uniform object. The first thing that came to mind was a puzzle; a section would be taken out of each ‘team model’ which, altogether, will create a complete cylindrical form.
The next thing to consider was which machine would be best to use; the 3D printer is the best for the job but one initial concern comes to mind; strength. Taking a section out of a cylinder with thin components (in this case the spacers in between each cylinder) will create weak points in the model. Because the majority of the weight is contained in the outer section of the model, depending on its weight, it may easily break.
CV curves were used to trace the curvature of the team model.
This was then revolved with a short span in order to create a section. The use of sections would be great in the sense that direct comparisons could be with any other team, giving a person great interactivity. However, the fragile nature of these parts made it hard to achieve this.
Starting with CV curves, a though came to position each curve in the same XYZ position then one-by-one, rotate each curve by an increasing angle (fraction of 360/13 = 27.69 degrees) creating a cylindrical form between the team-CV curves with 13 equal parts.
Lofting between each curve would create a harmonized object between the 13 team curves. Only one question remained, and this was the order in which the CV curves are placed. Alphabetical order was first considered, but this had no real relevance to the data in question. The geographical location was considered. This was however, also flawed as every order will result in end points that do not match (e.g. Stating from QLD, moving through to NSW, the ACT, then VIC and finally NZ; but QLD is not the closet to NZ). Therefore the ordering method used was determined by a number of factors; including geographical location, age of club and historic rivalries.
Once all the curves were ordered, a loft was created:
The “pointy” nature of the model would be fragile if printed and so a smoother solution was needed. This was a simple change from a linear loft to a cubic loft:
A few problems arose which could be challenging for the RP. There were certain sections on the sculpture that over lapped. This didn’t work in favour of the data, as the overlap hide underlying data.
Another problem was the paper-thin sections in some areas caused by a dramatic change in rank between two teams next to each other. This section would become quite fragile if printed and would more then likely break, when being removed form the printer. The option to scale to model up did come to mind however, the larger size may not suit the dimensions of the printer, not to mention the increased cost.
The great difference between neighbouring teams was the main cause for these problems. Changing certain vertex points could combat this problem (i.e. scaling points to increase thickness), but this will effect the data, changing it, therefore making it inaccurate.
The order of the teams needed to be changed. I realised the reasons for my current team order would be a little confusing to someone who doesn’t know much about rugby league. So I found an ordering solution which is self explanatory; order by overall ranking.
I added up the rankings for each team over the 11 year period and the team with the lowest score would be the one with the highest average ranking. the accumulated ranking is as follows:
- Brisbane Broncos: 48 - 1st
- St George Illawarra Dragons: 80 - 6th
- Newcastle Knights: 76 - 5th
- New Zealand Warriors: 101 - 11th
- North Queensland Cowboys: 120 - 13th
- Sydney City Roosters: 62 - 3rd
- Manly Sea Eagles: 98 - 10th
- Penrith Panthers: 110 - 12th
- Cronulla Sharks: 86 - 8th
- Canberra Raiders: 92 - 9th
- Melbourne Storm: 51 - 2nd
- Parramatta Eels: 66 - 4th
- Canterbury Bankstown Bulldogs: 83 - 7th
Teams which maintain high ranking over the 11 year period produced sections with a much greater length then those with poor rankings. Arranging them from highest to lowest will still create a problem where the two ends meet, so instead, the arrangement is as follows:
- The team with the highest ranking will be placed first.
- Next to them will be the teams with the second and third highest rankings.
- Next to these teams will be the teams with the fourth and fifth highest ranking
- So forth in this fashion for the rest of the teams.
This way no matter how you look at the sculpture, in whatever direction, the flow between one team to the next will always rise, then fall or vice a vera in a continual cycle.
Here is the result:
With this order, however, many fragile points were still created where thin points carried on between teams.
One thing I wanted to achieve through my model is the ability to compare not only one team to the others closest to it, but expand on this so one could be compared to every other team. The current model structure doesn’t allow for this, even though comparisons between teams isn’t that obvious, it can be achieved; via interaction.
By looking at the vertical structure one team produces and comparing this to another, assumptions can be generated about the two teams in question. The sense of touch can also be a valid tool, aiding the user’s interaction allowing for a greater sense of the different aspects of the sculpture.
With this in mind, the rotational order and placement of each team in relation to each other is something that is not absolutely necessary. The idea behind the sculpture is to consider individual parts separately, before making judgement about the sculpture as a whole.
The challenge was to develop an order structure between the CV curves which allows for the greatest strength as a tangible physical object. Order by overall height was unsuccessful, it did bring about a realisation; strength and stability is not obtained through connection of like parts, but rather differences are needed between such parts in order to create the greatest possible overall strength.
Instead of having an order of overall length (i.e. overall team ranking), all alteration between long and short CV curves creates a stable sculpture as the variation between a thin and thick section at a connection point (i.e. between 2 CV curves) allows the strength and weakness in both to cancel each other out. This was tested in Maya and was successful. There were still a few thin points, but these weren’t as bad as those in the previous models; they also appear strong enough to how when printed.
The order itself can also be considered as data driven, each CV curve of each team peaks at a certain point. By altering between tall and short peaks, the different curves are averaged out. This also mean that to distinguish an order (based on length) a person would need to examine all the different teams, thereby increasing interaction.
At this point the model was finished and just needed to be resized in Rhino in order for the model to be printed under the restrictions of the RP. There now was one final question to answer; how is the model going to be explained.
Since the sculpture represents football data, it seemed fitting to have the sculpture be presented like a trophy. The stand of the model contains the labels and the key. Support was needed to hold the model upright on the stand. In Rhino, using a Boolean difference between the sculpture and a cylinder a hole was created.
The reason for the cylindrical design is that if the hole was round, it will not be stable and will turn on its own axis. Support was also fitted between the stand and the model.
Once the model finished printing, I had the opportunity to go through process of extracting it from the 3D printer and finishing the surface.
First the excess powder is brushed off:
Then the remaining powder was blasted off:
It is then sprayed with water in order to strengthen the powder:
And finally baked in an oven:
A stand was created to house the labels needed:
Here are the images of the labels:
Final Product
Here is an annotated photo of the finished model:
And finally, here are the photos of the finished model:
References
http://www.nrl.com/Stats/PlayerStats/tabid/10253/default.aspx
http://www.nrlstats.com/index.cfm
http://www.rl1908.com/premiership.htm
Data Characteristics
Looking at each team, different trends start to appear. Some are quite unique, but others appear quite consistently between different teams. These data sets compare end of year rankings to overall home crowd sizes for that particular year between 1998 and 2008. The main reason for this is an assumption I have, that if a team performs consistently well, crowds will consistently attend games.
Brisbane Broncos
Season Ranking Total Crowds
1998 1 240878
1999 8 278791
2000 1 276082
2001 5 256780
2002 3 241575
2003 8 289023
2004 3 344007
2005 3 363923
2006 3 374494
2007 8 394415
2008 5 401108
Starting with the Brisbane Broncos, it can be seen from 1999, the crowd size begins to decrease and from 2003 onwards, the crowd size continuously increases. The same cannot be said about their rank, this does not follow in the same trend; even at the points where the team was ranked 1st, crowd numbers during that season where nowhere near the highest points. The team seems to have maintained a stable ranking in this 11 year period and the do not drift past the number 8 spot. In the sport of rugby league, teams strive to reach a top 8 position as this is the only way to move through to the finals (other teams being eliminated instantly). I may be said that as a result of this consistent performance, fans have stood by their team and crowd numbers are growing to some of the highest numbers ever recorded.
St George Illawarra Dragons
Season Ranking Total Crowds
1998 8 124728
1999 6 263692
2000 9 132884
2001 7 153772
2002 7 149389
2003 10 156911
2004 5 184058
2005 2 177358
2006 6 183182
2007 13 145908
2008 7 151154
The Dragons are one team which seam to more or less follow this assumption that team performance is influenced by crowd size. Nearly every instance where ranking improves, crowd numbers move forward with it. There is one instance, in 1999, where crowd number where at the highest point in 11 years. This suggests that something may have occurred during this year to bring about such high crowd numbers. In actual fact, this was the year the Dragons made it to the grand final match which may have had great influence in this data.
Newcastle Knights
Season Ranking Total Crowds
1998 2 243518
1999 7 250563
2000 3 243473
2001 3 228275
2002 2 222224
2003 7 217374
2004 9 207105
2005 15 221628
2006 4 262179
2007 15 190563
2008 9 224996
The Knights are a team which have both fluctuating crowd numbers and ranking and both of these data sets are inconsistent with each other. Like the Broncos, the Knights are the kind of team which maintain a steady rank. However, when they begin to drift outside the top 8, this is where we start to see the lower end of their crowd number and because of this assumption of rank affecting crowd size stands.
New Zealand Warriors
Season Ranking Total Crowds
1998 15 120478
1999 11 135265
2000 13 161042
2001 8 165940
2002 1 189082
2003 6 201536
2004 14 122025
2005 11 157212
2006 10 105946
2007 4 158699
2008 8 137325
This is the team which by far shows the greatest favourability for the assumption of rank affecting crowd size. Working down from one year to the next it can be seen that when rank improves, crowd numbers increase accept for 2003 where they reach and 11 year high. This may be due to their great performance in the pervious year where they came first, carrying on good performance in the following year.
North Queensland Cowboys
Season Ranking Total Crowds
1998 16 204981
1999 16 193296
2000 14 201771
2001 13 170732
2002 11 145600
2003 11 178709
2004 7 200357
2005 5 247075
2006 9 236175
2007 3 241262
2008 15 217228
This is a team which showed poor performance up until 2004 and after this point, not only does the rank increase, but the crowd numbers move up as well. Even in 2008 where they placed poorly, the Cowboys maintained a steady crowd size, demonstrating, like the Broncos, that crowd numbers are increasing.
Sydney Roosters
Season Ranking Total Crowds
1998 6 123104
1999 4 156380
2000 2 201426
2001 6 162961
2002 4 137389
2003 2 193161
2004 1 212053
2005 9 192681
2006 14 154151
2007 10 143421
2008 4 167642
Like the Warriors, the Roosters show great consistency between the two data sets. End of year rankings seems to definitely affect the crowd sizes for this club. This is supported by the fact that the points where the crowd numbers were at their highest, above 200 000, are when the team was place 1st and 2nd.
Manly Sea Eagles
Season Ranking Total Crowds
1998 10 126155
1999 13 215521
2000 12 169963
2001 10 133111
2002 9 112093
2003 14 124404
2004 13 134435
2005 8 182054
2006 5 189522
2007 2 181084
2008 2 164324
The assumption of ranking effecting crowd size seems to have a negative effect in the beginning of this data set. Their crowd numbers do seem to rise in the later part of the data but as their rank increases to an 11 year high point, crowd numbers begin to slowly fall. It could be said that the team does have support, but these fans do not like to go to home games. Only when the team’s performance starts to slip, the fans turn out to help give their team a push.
Penrith Panthers
Season Ranking Total Crowds
1998 14 111092
1999 10 149939
2000 5 186771
2001 14 131757
2002 12 132137
2003 1 212419
2004 4 211049
2005 10 186906
2006 12 138953
2007 16 144419
2008 12 130790
The Panther’s data shows that most of the time, the two data sets are consistent with each other. What’s more is that once again high ranking has draw the highest crowd numbers and both the performance and crowd numbers were maintained in the following year.
Cronulla Sharks
Season Ranking Total Crowds
1998 11 136444
1999 1 168295
2000 8 238974
2001 4 161513
2002 5 160890
2003 12 154738
2004 11 149865
2005 7 195371
2006 13 149859
2007 11 128302
2008 3 155581
The Sharks do show some consistency between the two datasets of rank and crowd numbers. What’s more interesting is fact the once again great performance in one year has lead to great crowd numbers in the following year. In 1999 the Sharks came 1st and in the following year, their crowd numbers hit an 11 year high with 238974. From this, it can be said after the team had a terrific season, the fans wanted to be there to try and help replicate this performance by supporting their team.
Canberra Raiders
Season Ranking Total Crowds
1998 7 118303
1999 9 144744
2000 4 158054
2001 11 124469
2002 8 125533
2003 4 160772
2004 8 133391
2005 14 148851
2006 7 137870
2007 14 138144
2008 6 142952
The Raiders don’t show much consistency between the two data sets. However like the Roosters and the Panthers, the greatest crowd numbers appear where rank was highest. So although there wasn’t any real consistency form one year to the next, higher ranking still produces greater rank maintaining come consistency in data correlation assumption.
Melbourne Storm
Season Ranking Total Crowds
1998 3 152688
1999 3 154827
2000 6 190092
2001 9 155745
2002 10 108693
2003 5 115509
2004 6 106583
2005 6 106770
2006 1 130426
2007 1 140527
2008 1 149686
Although crowd numbers are quite low for this team, the two data sets are still quite consistent with each other. The only place where this isn’t true is 1999, where rank decreases but crowd numbers hit an 11 year high. This does suggest that something may have occurred during this year to draw more crowds. In actual fact, 1999 was the year where Melbourne won the Grand Final, and this most likes would have been the cause for such high crowd numbers; fans wanting to witness, first hand, a repeat of results.
Parramatta Eels
Season Ranking Total Crowds
1998 4 136310
1999 2 174604
2000 7 186190
2001 1 271600
2002 6 169053
2003 9 135142
2004 12 141186
2005 1 195169
2006 8 175042
2007 5 169706
2008 11 162335
The Eels show some consistency between the two data sets, but once again, the point where crowd numbers where greatest was where rank was highest. This brings about the assumption that if a team is winning, the fans will come to games to support their team.
Canterbury Bulldogs
Season Ranking Total Crowds
1998 9 100358
1999 5 203450
2000 11 167489
2001 2 187557
2002 15 169919
2003 3 254280
2004 2 216382
2005 12 219431
2006 2 216489
2007 6 197852
2008 16 156564
The Bulldogs show a fair amount of consistency between the two data sets. Most of the lower ranks seem to generate the lowest crowd numbers. From 2006, you see a gradual slip of both crowd size and end of year ranking.
Looking at all these datasets it can be seen that in the majority of cases, end of year rankings will determine crowd size. Some teams demonstrate this quite consistently one a year to year basis. Others maintain this assumption due to the fact that when the team performs better then most in the comp (i.e. top 3 ranking), crowd number peak for these teams.
It can also be seen that team who do not reach this top rank do not achieve these high crowd numbers regularly. It is as if once it occurs, fan support rises, boosting the team to try repeat great performance. For example the Dragons, Raiders and Eagles are 3 teams without a number 1 spot in the past 11 years and their crowd numbers are lower then most other teams.
Comparing in the same region, it can be seen that although the Brisbane Broncos and North Queensland Cowboys have quite different rankings and crowd numbers, the distribution and progression of the crowd size is quite similar. The crowd numbers for both teams start off quite steady increasing or decreasing by small increments, and then suddenly, in 2004; crowd numbers for both teams start to dramatically increase. These are the only tow teams with this same structure which would suggest a great increase in fan support in the state over the past few years.
The Melbourne Storm are a team which stand out from the others. Unlike any other team, they have consistently high ranking but some of the lowest crowd numbers out of any other team. For a team which has scored more 1st rankings in the past 11 year than any other team this does suggest that something is wrong. Fact of matter is that in Victoria, AFL is the sport of choice, dominating the state, leaving room for a lot less supports of rugby league. The NZ Warriors is another team which demonstrates this. Outside influence of rugby union once again drains the fan base for rugby league.
Low crowd numbers could be caused by something else. The teams with the lowest crowd numbers include the Canberra Raiders, The Melbourne Storm and the NZ Warriors. These teams have no competition close by (geographically) which brings about further assumptions. Supporters of these teams may not be as fierce as others in NSW and OLD and many of them don’t have to constantly deal with fans that support other teams. It could also be that when these teams have how games, fans of the opposing team are not willing to make an extended trip to a game interstate, or in another country.
Design Rationale
The concept for the data model began with 2 datasets; end of year ranking and crowd numbers for all the teams in the NRL from 1998 to 2008. The reason for the 11 year period is the fact that before 1998 there were complications in the league; some teams were folding others was merging. Another reason for this short time period is to take a more modern look at the game. Looking at a sport now compared to 20 years ago, no matter what the sport is, the sport would have always changed as a result of greater support, funding, and human abilities.
Running through and comparing these two datasets against time allowed for one team to stand out, the NZ Warriors; where a change in one dataset was consistent with the change in the other, moving though one year to the next.
The hardest thing was now designing a data-mapping technique which allowed each dataset to be dependent on one another (i.e. if one of the two datasets was removed, the shape and form of the remaining data set will change).
Starting with simples shapes, a sculpture was created. Using cylinders to represent each year, each cylinder was stacked on top of the other representing an annual increase with each new cylinder. The net 2 attributes affected the shape of each cylinder. Crowd size determines the radius. The larger the crowd size, the greater the radius. End of year rank determined the height/thickness of each cylinder. The greater the rank, the thicker the cylinder. A scaling system was created so that in Maya, cylindrical thickness values could easily and accurately be changed. The scale is as follows:
1st – 1.6
2nd – 1.5
3rd – 1.4
4th – 1.3
5th – 1.2
6th – 1.1
7th – 1
8th – 0.9
9th – 0.8
10th – 0.7
11th – 0.6
12th – 0.5
13th – 0.4
14th – 0.3
15th – 0.2
16th – 0.1
In Maya it became hard to determine between cylinders (i.e. one year from the next) and so spacers were created between each cylinder.
After looking at the sculpture that was created, I found the dataset to be quite small, and not much meaning could be formed from a single team. Comparisons were needed between other teams to allow for greater reliability (i.e. finding similarities between different teams in order to support a given meaning). The reason for starting with only one team’s datasets was because it was the one which showed the most consistency between the two datasets from one year to the next. Now the question that was why are they the most consistent team? Sculptures were then created for every other team in the competition who has competed in every season for the past 11 years. Other teams were left out because they could cause irregularities in the final sculpture in all the team-sculptures were combined.
The order of these sculptures from left to right is bulldogs, eels, storm, raiders, sharks, panthers, eagles, roosters, cowboys, warriors, knights, dragons, broncos.
Each sculpture seemed to have its own unique shape and told its own story, but side by side, other patterns started to appear (discussed above).
Now the challenge became the integration of all 13 models into one uniform object. The first thing that came to mind was a puzzle; a section would be taken out of each ‘team model’ which, altogether, will create a complete cylindrical form.
The next thing to consider was which machine would be best to use; the 3D printer is the best for the job but one initial concern comes to mind; strength. Taking a section out of a cylinder with thin components (in this case the spacers in between each cylinder) will create weak points in the model. Because the majority of the weight is contained in the outer section of the model, depending on its weight, it may easily break.
CV curves were used to trace the curvature of the team model.
This was then revolved with a short span in order to create a section. The use of sections would be great in the sense that direct comparisons could be with any other team, giving a person great interactivity. However, the fragile nature of these parts made it hard to achieve this.
Starting with CV curves, a though came to position each curve in the same XYZ position then one-by-one, rotate each curve by an increasing angle (fraction of 360/13 = 27.69 degrees) creating a cylindrical form between the team-CV curves with 13 equal parts.
Lofting between each curve would create a harmonized object between the 13 team curves. Only one question remained, and this was the order in which the CV curves are placed. Alphabetical order was first considered, but this had no real relevance to the data in question. The geographical location was considered. This was however, also flawed as every order will result in end points that do not match (e.g. Stating from QLD, moving through to NSW, the ACT, then VIC and finally NZ; but QLD is not the closet to NZ). Therefore the ordering method used was determined by a number of factors; including geographical location, age of club and historic rivalries.
Once all the curves were ordered, a loft was created:
The “pointy” nature of the model would be fragile if printed and so a smoother solution was needed. This was a simple change from a linear loft to a cubic loft:
A few problems arose which could be challenging for the RP. There were certain sections on the sculpture that over lapped. This didn’t work in favour of the data, as the overlap hide underlying data.
Another problem was the paper-thin sections in some areas caused by a dramatic change in rank between two teams next to each other. This section would become quite fragile if printed and would more then likely break, when being removed form the printer. The option to scale to model up did come to mind however, the larger size may not suit the dimensions of the printer, not to mention the increased cost.
The great difference between neighbouring teams was the main cause for these problems. Changing certain vertex points could combat this problem (i.e. scaling points to increase thickness), but this will effect the data, changing it, therefore making it inaccurate.
The order of the teams needed to be changed. I realised the reasons for my current team order would be a little confusing to someone who doesn’t know much about rugby league. So I found an ordering solution which is self explanatory; order by overall ranking.
I added up the rankings for each team over the 11 year period and the team with the lowest score would be the one with the highest average ranking. the accumulated ranking is as follows:
- Brisbane Broncos: 48 - 1st
- St George Illawarra Dragons: 80 - 6th
- Newcastle Knights: 76 - 5th
- New Zealand Warriors: 101 - 11th
- North Queensland Cowboys: 120 - 13th
- Sydney City Roosters: 62 - 3rd
- Manly Sea Eagles: 98 - 10th
- Penrith Panthers: 110 - 12th
- Cronulla Sharks: 86 - 8th
- Canberra Raiders: 92 - 9th
- Melbourne Storm: 51 - 2nd
- Parramatta Eels: 66 - 4th
- Canterbury Bankstown Bulldogs: 83 - 7th
Teams which maintain high ranking over the 11 year period produced sections with a much greater length then those with poor rankings. Arranging them from highest to lowest will still create a problem where the two ends meet, so instead, the arrangement is as follows:
- The team with the highest ranking will be placed first.
- Next to them will be the teams with the second and third highest rankings.
- Next to these teams will be the teams with the fourth and fifth highest ranking
- So forth in this fashion for the rest of the teams.
This way no matter how you look at the sculpture, in whatever direction, the flow between one team to the next will always rise, then fall or vice a vera in a continual cycle.
Here is the result:
With this order, however, many fragile points were still created where thin points carried on between teams.
One thing I wanted to achieve through my model is the ability to compare not only one team to the others closest to it, but expand on this so one could be compared to every other team. The current model structure doesn’t allow for this, even though comparisons between teams isn’t that obvious, it can be achieved; via interaction.
By looking at the vertical structure one team produces and comparing this to another, assumptions can be generated about the two teams in question. The sense of touch can also be a valid tool, aiding the user’s interaction allowing for a greater sense of the different aspects of the sculpture.
With this in mind, the rotational order and placement of each team in relation to each other is something that is not absolutely necessary. The idea behind the sculpture is to consider individual parts separately, before making judgement about the sculpture as a whole.
The challenge was to develop an order structure between the CV curves which allows for the greatest strength as a tangible physical object. Order by overall height was unsuccessful, it did bring about a realisation; strength and stability is not obtained through connection of like parts, but rather differences are needed between such parts in order to create the greatest possible overall strength.
Instead of having an order of overall length (i.e. overall team ranking), all alteration between long and short CV curves creates a stable sculpture as the variation between a thin and thick section at a connection point (i.e. between 2 CV curves) allows the strength and weakness in both to cancel each other out. This was tested in Maya and was successful. There were still a few thin points, but these weren’t as bad as those in the previous models; they also appear strong enough to how when printed.
The order itself can also be considered as data driven, each CV curve of each team peaks at a certain point. By altering between tall and short peaks, the different curves are averaged out. This also mean that to distinguish an order (based on length) a person would need to examine all the different teams, thereby increasing interaction.
At this point the model was finished and just needed to be resized in Rhino in order for the model to be printed under the restrictions of the RP. There now was one final question to answer; how is the model going to be explained.
Since the sculpture represents football data, it seemed fitting to have the sculpture be presented like a trophy. The stand of the model contains the labels and the key. Support was needed to hold the model upright on the stand. In Rhino, using a Boolean difference between the sculpture and a cylinder a hole was created.
The reason for the cylindrical design is that if the hole was round, it will not be stable and will turn on its own axis. Support was also fitted between the stand and the model.
Once the model finished printing, I had the opportunity to go through process of extracting it from the 3D printer and finishing the surface.
First the excess powder is brushed off:
Then the remaining powder was blasted off:
It is then sprayed with water in order to strengthen the powder:
And finally baked in an oven:
A stand was created to house the labels needed:
Here are the images of the labels:
Final Product
Here is an annotated photo of the finished model:
And finally, here are the photos of the finished model:
References
http://www.nrl.com/Stats/PlayerStats/tabid/10253/default.aspx
http://www.nrlstats.com/index.cfm
http://www.rl1908.com/premiership.htm
Comment
About
© 2010 Created by Andrew on Ning. Create a Ning Network!
You need to be a member of infostudio 2009 to add comments!
Join this Ning Network