This small multiple set of scatter plots shows the points vs opponent points for each NBA team in 2013.
The small multiples are sorted by Pearson Correlation, (shownÂ on the top right),
The Knicks final score is least dependent on what their opponent is doing.
On the other end of the spectrum, the Raptor are happy to score high or low, depending on the opponent.
The motivated viewer sees that the highest scoring game was 145-130, Rockets over Lakers.
The Bulls favor low scoring games, except for two.
The most common margin of victory in 2013 College basketball was 3, followed by 4,5 and 2.
In the NBA the most common margin in 2013 was 7.
Why the difference?
The most common score in Canadian Football is 23 – 20.
Ties are rare and have occurred at 0, 39, 44 and 45.
The highest scoring game was 54 – 51.
This scatter plot shows the team points vs opponent points for all NBA games in 2013.
One sees quickly that the highest scoring game was 145 to 130ish.
The strange brain shaped pattern shows that NBA teams avoid very close scores: in fact a margin of 7 was most common in 2013.
If the teams’ scores were independent we would expect a circle rather than a brain.
I like the gigantic transparent icons because the viewer observers the gray scale directly.
Below is a cleaner version which avoids overlapping icons.
One sees more clearly that there are no ties in the NBA and although, ‘we knew that’, mapping this knowledge to the graphic roots the viewer and gives confidence for further insights.
It seems not inappropriate to spread out these final scores visually – what if that last shot went in ?
What do you think?
The final score (points vs opponent points) for every NFL game 1989 – 2013.
This scatter plot works as a ‘heat map.’
We see quickly that:
- A total of 1 never happens (it is impossible by NFL scoring rules)
- A total of 2 is rare (a safety)
- A total of 4, while possible, has never happened.
- The most common score is around 23, or somewhere in there.
- Ties happen and are rare.
It could probably be improved with a better gradation scheme.
This colorfully stacked histogram shows the distribution of points minus opponent points (the margin) for all NBA games in 2013.
I like it because one sees quickly that:
- Seven was the most common final score margin, followed by 6, 5, 4, and then 8 and 9 were equally common;
- A margin of 1 was as common as 11.
The colors are arguably gratuitous as individual teams are hard to distinguish.
One does see a purple bump to the right; are we learning anything?
This more austere version chooses one color and shows each game once.
To build histograms like this yourself, start with the SDQL; margin @ season=2013 and team
It was most common for an MLB stater to last 6 innings in 2013. A complete game of 27 outs is unusual.
I’m just learning about matplotlib’s coloring rules. Â Here I am using: plt.get_cmap(‘gist_rainbow’) and might have to look a little harder there.
I like this histogram because we see from a distance that Giants starters most often go 7 innings, followed by 6 and 5.
I like this because we see quickly that Lincecum and Petit had complete games and for 8 complete innings we can just about read off: 2 for Lincecum; 3 for Bumgarner, 3 for Cain; and 1 for Vogelsong.
To improve it; show each game as a block, rather than bars of some length. This reinforces the idea of a unit and we can count small stacks.
Also, I had trouble with the colors in the green-blue. Is that me?
This beautiful small multiple set shows how many outs the starter got in 2013. The Tigers starters got 21 outs 49 times during 2013 and the Braves starters lasted 18 outs also 49 times.
The finger-like structure is due to the starting going a complete number of innings.
The bands in the histograms represent different starters; only experts can guess who is blue.
The Nationals like their starters to finish the inning; Cleveland and the Angles less so.
It looks like there are a lot of stories in there.
See anything interesting?
This Matplotlib generated box-plot small multiple shows the distribution for each NBA team’s fantasy points in 2013.
I like it because one distinguishes quickly a super-star driven team with a steep profile from a more balanced team with a flatter profile.
It could be improved by featuring the data rather than the constructed box.
The Grizzlies stand out in that their top three have very similar fantasy point production and then quite a drop off.
This image was generated dynamically using the small multiple engine at http://SportsDatabase.com
The SDQL is simple: FP @ team and name and season = 2013