Weighting Correlation Co-Efficients to measure performance

[olimajor123]

Hi,


I have some interesting data based on soccer that I am trying to use to measure performance based on the correlation coefficients.

I have data for each player in the English Premier League that shows how many goals they have scored and with this range of other data that contributes to this, see example below:

[TABLE="width: 805"]
<tbody>[TR]
[TD]Name[/TD]
[TD]Team[/TD]
[TD]Goals[/TD]
[TD]Big Chances[/TD]
[TD]Goal Attempts[/TD]
[TD]Shots - Inside Box[/TD]
[TD]Shots - Six Yard Box[/TD]
[TD]Shots On Target[/TD]
[TD]Time Played[/TD]
[TD]Touches - Penalty Area[/TD]
[TD]Minutes Per Chance[/TD]
[/TR]
[TR]
[TD]Ibrahimovic[/TD]
[TD]MUN[/TD]
[TD]15[/TD]
[TD]16[/TD]
[TD]95[/TD]
[TD]67[/TD]
[TD]11[/TD]
[TD]39[/TD]
[TD]2186[/TD]
[TD]209[/TD]
[TD]23[/TD]
[/TR]
</tbody>[/TABLE]


I have produced a correlation co-efficient table based on this info to see the link between goals scored and all the other factors to give coefficient values for each one, see below:

[TABLE="width: 580"]
<tbody>[TR]
[TD][/TD]
[TD]Big Chances[/TD]
[TD]Goal Attempts[/TD]
[TD]Shots - Inside Box[/TD]
[TD]Shots - Six Yard Box[/TD]
[TD]Shots On Target[/TD]
[TD]Time Played[/TD]
[TD]Touches - Penalty Area[/TD]
[TD]Minutes Per Chance[/TD]
[/TR]
[TR]
[TD]Goals[/TD]
[TD]0.88[/TD]
[TD]0.79[/TD]
[TD]0.86[/TD]
[TD]0.64[/TD]
[TD]0.87[/TD]
[TD]0.33[/TD]
[TD]0.80[/TD]
[TD]-0.34[/TD]
[/TR]
</tbody>[/TABLE]



What I want to do is make a rough formula to predict how many goals the player 'should have' scored based on the data (chances, shots etc) using the co-efficients to weight the factors towards this. IE if big chances has the highest co-efficient it should be weighted more in the formula. The final goal of this would be to measure who has over and underperformed in scoring goals given the data provided.

I hope this makes sense, can anyone help?

 

[DRSteele]

Unless I'm way off base, I fear that it's not as simple as correlation coefficients. You need to conduct a multiple Regression Analysis. This will produce the Regression Equation for predicting Goals, the other parameters and the error. E.g., Goals = constant + Chances + Attempts + Inside + Six +Target + Played + Touches + Minutes + standarderrror

Personally, I find most of these concepts baffling and trying to make sense of it all makes me frown. But watching this video on youtube from ExcelIsFun might just get us on the right track. https://www.youtube.com/watch?v=1oBXXsxcp5k Also, reading through this might help too: Multiple Regression | Real Statistics Using Excel

That ExcelIsFun channel is a tremendous resource; it has dozens of videos related to statistics and their application in Excel. Perhaps you might find it educational.
<strike></strike>
Excel has an and-in that will do most of the heavy work: Analysis ToolPak<strike></strike>

 

[olimajor123]

Unless I'm way off base, I fear that it's not as simple as correlation coefficients. You need to conduct a multiple Regression Analysis. This will produce the Regression Equation for predicting Goals, the other parameters and the error. E.g., Goals = constant + Chances + Attempts + Inside + Six +Target + Played + Touches + Minutes + standarderrror

Personally, I find most of these concepts baffling and trying to make sense of it all makes me frown. But watching this video on youtube from ExcelIsFun might just get us on the right track. https://www.youtube.com/watch?v=1oBXXsxcp5k Also, reading through this might help too: Multiple Regression | Real Statistics Using Excel

That ExcelIsFun channel is a tremendous resource; it has dozens of videos related to statistics and their application in Excel. Perhaps you might find it educational.
<strike></strike>
Excel has an and-in that will do most of the heavy work: Analysis ToolPak<strike></strike>

Thanks - this was really really helpful and I have managed to do this now. I trimmed it down and just used the data that had the highest co-efficient as would be the most reliable and then used the regression tool shown in the video.

Thanks a lot

 

[DRSteele]

Spendid. I'm glad your journey wasn't a wild goose chase.