Survey banner
Share your feedback on the Dataiku documentation with this 5 min survey. Thanks!

# Conundrum 23: Visualising Victors

Community Manager
###### Conundrum 23: Visualising Victors

Welcome to Conundrum 23!

This week letโs take a look at some super light data prep and visualisation in the wonderful world of chess.

Attached is data about a collection of chess games - included is the ID of the players, the victor of each game, and a few other statistics.

Can you display the top 10 players by raw win:loss ratio and display their prowess visually?

Good luck!

I hope I helped! Do you Know that if I was Useful to you or Did something Outstanding you can Show your appreciation by giving me a KUDOS?

Looking for more resources to help you use DSS effectively and upskill your knowledge? Check out these great resources: Dataiku Academy | Documentation | Knowledge Base

2 Replies

@MichaelG and others who are reading.

Thanks for this week's conundrum.

This conundrum is using an ambiguous term, or a term that I do not understand how to calculate.

`raw win:loss ratio`

If a player played 1 game and lost 0 as many of the participants in this data did.  That produces an undefined value for the ratio win/loss, often displayed as a positive Infinity.  There are 4381 players who have no losses in the data. (if you drop draws.) All would be calculated as an infinite win/loss ratio.

However, then how do we pick a top 10 from this list?  All have the same infinite score. We might want to celebrate a top 10 pick based on the number of wins.   So showing the player who had 24 wins with no loss might get the #1 position.  A then the player with 18 wins and no loss...

However what about a player that has 45 wins and 1 loss?  They made 46 attempts and have a ratio of 45/1 which is lower than the infinity scored by a player who had 1 win and 0 losses.  I would think that the player having 45 wins and 1 loss is likely a much better player than the "one-win wonders" in our dataset.

Can anyone point me to a place that produces a calculation to deal with this kind of issue?

@MichaelG can you clarify what you intended by the

`top 10 extract the top players by raw win:loss ratio`

P.S. What do folks think about dealing with the 950 games that are "draws"?  At this point in time, I've dropped those games because they are neither a win nor a loss.  We could also give each player .5 wins and .5 losses for the draw.

--Tom
Community Manager
Author

Hey @tgb417

Thanks for the question/input!

I agree those players with a infinite win/loss ratio present a problem for how I formulated the conundrum - perhaps raw wins would be a better metric. But then that would bias in favour of those plays who played a lot of games - since 10 wins 5 losses would rank lower than 20 wins and 50 losses.

Perhaps a combined score that requires a given number of games played and then awards a value based on the number of wins less the number of losses? Any other ideas?

On what exactly I intended by that I'm afraid that was just a mistype on my part - I meant what it now says:

`top 10 players by raw win:loss ratio`

Thanks for pointing that out!

I hope I helped! Do you Know that if I was Useful to you or Did something Outstanding you can Show your appreciation by giving me a KUDOS?

Looking for more resources to help you use DSS effectively and upskill your knowledge? Check out these great resources: Dataiku Academy | Documentation | Knowledge Base