Is There a Simple Way to Chart Bowling Scores for a Large Group of Players?

[Moderation]

Hi and thanks for viewing this. I am looking for as many ideas as possible so if you see a reply please do not hesitate to reply also.

To make this example easy I am going to explain my isse using group of bowlers as an example.

OK let's say I have 5000 bowlers and they are all on one team. They give me their scores for each week and I input these scores they give me. I want to show a way to graph/chart these scores on ONE SCREEN so they can view them. HERE'S THE MOST IMPORTANT PART they need to be able to see any significant change in their scores as time progresses QUICKLY and EASILY. Meaning these people are not chartists or math maticians so I can't use something that is difficult to decipher. The scores will have weekly sums going back several years. So the screen will have lots of numbers. I am fine with replacing the scores with anything to show the significant change in value, but again it needs to be easily read when I look look over ALL the bowlers data and simply scan looking for any significant change going back several years.

 

[story645]

OK let's say I have 5000 bowlers and they are all on one team. They give me their scores for each week and I input these scores they give me. I want to show a way to graph/chart these scores on ONE SCREEN so they can view them. HERE'S THE MOST IMPORTANT PART they need to be able to see any significant change in their scores as time progresses QUICKLY and EASILY.

How robust does the system need to be and how skilled are you as a coder?

You can implement this in excel or access + maybe a vb based gui. I like python/sqlalchemy/postgres + matplotlib for display/phpdbadmin for views (and pyQT or pyGTK for a gui) but that could be overkill for your task.

Simple line graphs will show significant changes well enough and don't require much knowledge.

 

[Moderation]

Hi and thanks for your idea. I am open to any suggestion. I think a line chart might be pretty busy (5000 charts on one screen shot) and hard to decipher, but I appreciate your replying.

 

[story645]

Hi and thanks for your idea. I am open to any suggestion. I think a line chart might be pretty busy (5000 charts on one screen shot) and hard to decipher, but I appreciate your replying.

I meant you generate charts on request, which is why you'd probably want to cobble together a GUI.
If you've got to have all 5000 on a screen, just throw however many weeks numbers and use arrows (up if the scores have improved, down if they haven't, equal if it's stable) based on whatever metric you want. Have a few cells: week based changes (how this week compares to last), month based (how this week compares to the month and how this month compares to last month), and year based (how this month compares to the year, and how this year stacks up to last.) Adjust as needed and ask your users which metrics they really care about. On the backend you've got to do some real statistics (I think a t-test works for this, but not sure), but just report significance through a graphic.

 

[Moderation]

Thanks Story645.

How about if the group wants to find those individuals that have had significant score improvements or losses? Meaning I look at 5000 players and I want to see easily and quickly those that have moved up or down more than the others. I thought about just using raw numbers or percentages, but with 5000 it's just too busy. I need a way to make it simple for laymen to just look/scan and find those anomalies compared to the rest broken down in weekly periods (many years worth).

 

[story645]

How about if the group wants to find those individuals that have had significant score improvements or losses? Meaning I look at 5000 players and I want to see easily and quickly those that have moved up or down more than the others.

Flags! Put those bowlers in a different color (diff colors for up and down if you'd like) or push them to the first 20 positions or throw on a star by their names or use two arrows or different colored arrows for the diff levels (this may be too busy/require people to remember things, so not my fave option) or whatever else you think will make them distinctive. You''re just changing your significance thresholds on the backends.

I like mangaupdates stats implementation (it's totally what's inspiring my answer).