Alright, now we have all the data we need in one dataframe. To make this code work, I assume you ran the code from Part 1. We need the dataframe big.tab.
All the data presented here is based on the data from 18/10/2012. You can run an analysis with the actual data or I can do it at some point later in the season.
Let’s plot some stuff. How about the old german saying about soccer “Geld schießt keine Tore” (Money doesn’t score goals)? Let’s look into this.
plot(big.tab$Value, big.tab$Goals.for, type = “n”, axes = F, xlab = “Value”, ylab = “Goals”)
text(x = big.tab$Value, y = big.tab$Goals.for, labels = big.tab$Team, cex = 0.7, col = “#65656599″)
axis(side = 1)
axis(side = 2)
We get this… (clickable)
Sorry for the overlapping team names. But you get the gist: It looks like the value of a team covaries with the number of goals for that team.
Now we add a regression line. This means, we predict the number of goals for one team by the value of that team. We also add a Pearson correlation coefficient (r) and its associated p value in the subtitle of the plot. We get this…
How do we interpret this? There are several conclusions that could be drawn.
(1) The value of a team in the british Premier League is reliably correlated with the number of goals that team scored in the championship so far (after 7 games). Beware: Correlation does not imply causation
(2) The “best guess” of predicting the number of goals by the value of a team is visualized by the dashed red line in the second plot. This means that there are teams who “over-perform” and “under-perform” in relation to their value. FC Fulham, for example, shot way “too many” goals given its value. FC Liverpool, on the other hand, should have shot more goals, because they are under the red line.
(3) One could infer from this plot that it is quite difficult for very valuable teams (e.g., ManU, ManCity and the FC Chelsea) to over-perform since the regression line is so steadily rising. So, they have to score many many goals to outperform their level on the regression line.
By the way: This also works quite good for the value of a team and the points they achieved in the championship (win = 3 points, draw = 1 point).
In the next post
, I will do some more analyses and plots with this dataset. And I will try to compare different european championships.
To leave a comment
for the author, please follow the link and comment on his blog: Rcrastinate
offers daily e-mail updates
news and tutorials
on topics such as: visualization (ggplot2
), programming (RStudio
, Web Scraping
) statistics (regression
, time series
) and more...
If you got this far, why not subscribe for updates
from the site? Choose your flavor: e-mail
, or facebook