SportsData Challenge

DataViz challenge for the sports analytics community - aiming to improve visual creativity and standards across the community via healthy competition #OpenToAll

[SportsData Challenge @SportsDataChal]

EPL:Weighted Minutes for Academy Graduates. Not a single @LFC Academy graduate in the squad today having sold a lot of there talent

[SportsData Challenge @SportsDataChal]

@deepxg @ProformAFC Will flexclust help with the 2nd bit?

[SportsData Challenge @SportsDataChal]

@deepxg @ProformAFC That's a great suggestion... thanks

I was also hoping to identify generic 'clusters' of situations which would help with finding matching positional setups but also see frequency/style of a team

[SportsData Challenge @SportsDataChal]

Is there any way to do a k-means cluster analysis with a full matrix for each observation? #rstats

[SportsData Challenge @SportsDataChal]

@neilcharles_uk Was it more like this?
https://t.co/QDI3PJx83P

[SportsData Challenge @SportsDataChal]

Close of the day progress...

1. Happy with how it is matching
2. Video Analyst enters time of situation he wants to find more of
3. Run script
4. Exported to XML is working into SportsCode
5. Double click the row and watch all the clips

#rstats #analytics
example below

[SportsData Challenge @SportsDataChal]

@andrewjwant Yeah, this is the area where I would love to improve it. I simply don't know if there are better alternatives to cor(). I am very happy with the eyeball test of it after the last 20 minutes improvements

[SportsData Challenge @SportsDataChal]

@andrewjwant I spoke to soon!!

[SportsData Challenge @SportsDataChal]

@andrewjwant Thanks.. just made some nice progress over the last 20 mins as well

[SportsData Challenge @SportsDataChal]

@andrewjwant Once tracking data is loaded.. It currently takes less than 60 seconds to complete everything including matrix comparisons for positions of team1 and team2 at second intervals

[SportsData Challenge @SportsDataChal]

@andrewjwant Thanks for suggestion... I am not sure, will investigate

[SportsData Challenge @SportsDataChal]

@PenaltyKickStat but yes... it has lots larger utility in other things you mentioned

[SportsData Challenge @SportsDataChal]

@PenaltyKickStat exactly... I am looking at the automation of match preparation processes to start with. So.. for opposition 'x' I have noticed something good .. pattern 'y' ... enter the time code.. hit run and then all occurrences are timecoded.

[SportsData Challenge @SportsDataChal]

@PenaltyKickStat I am a novice! think of an idea and then magpie, google answers and eventually work it out :)

Yep, number = amount of players in that bucket.. indeed just for one team. I have it working for both teams and checking the correlation but not happy with the matching for 2 teams

[SportsData Challenge @SportsDataChal]

A few bugs cleared..

It's matching pretty well and works as a tool... I need to work out export to xml to import to SportsCode which I have done before so shouldn't be an issue

Top left is the situation that was searched others are the matched results

Feedback?

#rstats

[SportsData Challenge @SportsDataChal]

Updated the GitHub -> https://t.co/GDVNtHWhLm

I have the positional matrices for each second of the game.. I can enter a frame and get the closest matching second in the game - see pic.

I am using cor() in R for this... any suggestions of better ways? #rstats

[SportsData Challenge @SportsDataChal]

Wanted to display the matrices without the 0 to make it easier to view

[SportsData Challenge @SportsDataChal]

Select a frame in time and fill the matrices for team 1 and team 2 for that frame

[SportsData Challenge @SportsDataChal]

@PenaltyKickStat haven't tried before

[SportsData Challenge @SportsDataChal]

will be updating this --> https://t.co/HN6gHFG2hu

calculated buckets, created empty matrics and got spacing right

[SportsData Challenge @SportsDataChal]

would be grateful for suggestions for #stuff2