lunedì 31 ottobre 2016

3 point shot trend (1980-2016)

Since the introduction of the 3 point shot, back in 1979/80 season, its percentage distribution changed over time.

Starting from just the 3,1 percent of the total field goal attempts, this percentage is still increasing nowdays (29,1%).

You could say: "Hey, what about the mid 90's peak"?
And I would say: "For just 3 seasons, the NBA decided to set the 3point line at 22 feet (no matter top of the key or corners) and so, the percentage of total field goal hit a spike".

You can notice as well that the accuracy of 3 point shot has definitely improved through time (orange to blue): around 25% in the early 80's opposed to a "stable" 35+ % since late 90's

Bryant + Duncan + Garnett = 11 seasons by 76ers

It's fun (or sad, as you prefer) to see how much we missed since Bryant, Duncan and Garnett decided to retire after last NBA season.

To better understand the volume of their career numbers, I've decided to compare these numbers with an entire team.
Taking the 76ers just for numerical analysis purpose, it takes 11 complete roster seasons to match these three retired legends numbers.

domenica 30 ottobre 2016

No country for "Big" men

Maybe it's too early in the NBA season but if you look at the top50 scorers right now (at most 3 games), 47 of them have already attempted a 3 point shot.

Just 3 of them, Valanciunas, Whiteside and Nurkic are playing like old school Big men.

sabato 29 ottobre 2016

Best Season Challenge episode 3: Kobe Bryant vs Tracy McGrady

Since it seems that you guys are lovin' it, I've just finished this episode of #BestSeasonChallenge.

For this challenge, it's time to compare Kobe Bryant (2005/06) and Tracy McGrady (2002/03), two of the deadliest and most terrific scorers of the new millenium.

giovedì 27 ottobre 2016

Belinelli 2013/14 championship stats

During 2013/14 season, Marco Belinelli emerged as one of the deadliest shooters in the league and he also won the 3-point Contest during the All Star Game.

Headed to a promising team like Charlotte, will Belinelli go back to those numbers?

Here is a simple dataviz representing his shot distribution over the floor during 2013/14 season back in his Spurs days.

BTW: just for italian friends, this my Belinelli Day resume; I gave him a custom made t-shirt in his Honor (see pic)

mercoledì 26 ottobre 2016

NBA Salaries vs Player Efficieny Ratings

How much is a NBA player valuable?

With the new salary cap, how many superstars are currently underpaid?

You said Steph Curry? Probably you're right... check his salary rank!

Well, why don't you find the overpaid?

2015: The Rise of the Splash Brothers

With the arrival of Kevin Durant, the Golden State Warriors have added one of the deadliest scorer in the history of the NBA and their fire power will be insane.
KD will join a special duo, known as the Splash Brothers that last year was a sort of nightmare for every opponent team.

Despite being on the spotlight for most part of 2016, thanks to the Warriors historical 73-9 all time record, Stephen Curry and Klay Thompson probably started their career as Splash Brothers back in the 2015 when they were headed to their first NBA title.

Here's a look to the 2015 solar year with multiple and interactive visualizations that you can filter to see different aspects of this duo. Every viz is connected to the other ones, so basically, you have multiple options to explore their stats.

lunedì 24 ottobre 2016

NBA Preseason Rank based on Advanced Stats

Just one more day to the NBA tipoff.

After this preseason, is it possible to understand any of the NBA teams trend both on the defensive and offensive end?
Yes or no? You decide!

I've just discovered some interesting facts:
  • 4 defensive-minded teams: Hawks, Pistons, Spurs and Timberwolves (Tom Thibodeau anyone?)
  • Rockets and Knicks could be really prolific on the offense, but what about defense?
  • Where are the Cavs? Maybe too many players to watch&try... they're too low ranked
  • Could this be a pivotal year for the Celtics with the addition of Al Horford?
  • Miami Heat exploit is just a preseason anomaly?
What about you?

I've calculated a rank for each of the four main stats involved in this analysis and then I've tried to "summerize" a sort of cumulative rank based on the average of these 4 previous ranks.

Golden State gets the spotlight. Are you surprised?

Dwyane Wade departure from Miami

This summer, Dwyane Wade decided to leave Miami after 13 years with the Heat in order to play for the Chicago Bulls, his hometown team.

Despite not being in his prime anymore, he can bring his championship experience and veteran leadership to the Windy City.

Since the arrival of LeBron James in Miami, Wade had to to adjust his game to be the second offensive option and probably he will have to do the same in Chicago, sharing the ball with rising star Jimmy Butler.

Check his number starting from the Big3 era.

Every visualization in the dashboard is connected, so you can filter across data: for example you can start clicking one spot on the map to see just its details in the right charts.

United States Debt

For this week MakeOver Monday edition, I had to produce a dataviz strating from the smallest dataset I've ever used:
2 rows with just one value.

You may say it's really simple... well it's not!
If you have small data, you can't make mistakes.

So I opted for home-made unichart and a donut chart to represent %'s values.
Here's the viz:

venerdì 21 ottobre 2016

Larry Bird & Magic Johnson : Eternal Rivalry... even with numbers

Since Jordan vs James comparison has got a lot of love, I've decided to make another comparison... just a little bit more nostalgic.
And talking about nostalgia, what about 80's?

And if I say 80's, we ALL know what's coming: Bird vs Magic!

No need to tell you about their rivalry and all the Celtics-Lakers battles, but once again, here comes a funny fact.
In my humble opinion, Bird a Magic best season (stats wise) has been the same: 1986/87.

You should already know how it works: use and interact with bars to see how many games these two legends finished with at least X points, Y rebounds and Z assists.

Remember they were double/double machines... Find out how many triple/double they recorded.

giovedì 20 ottobre 2016

Jordan & James: 2 Numbers 23, but the magic number is 25

Disclaimer: I don't want to create any sort of flame because there's already plenty of it on the web!

Waiting for the start of the NBA season, I was taking a look to some Micheal Jordan stats when I realized that his best statistical season could have been the 1988/89.
Back then, he was 25.
Thinking about his Airness, and his 23 jersey number, I was wondering:
"Hey, what about LeBron James best season?"

For some reasons, LBJ 2009/10 season was his best one in terms of stats.
Guess how old he was?
Correct: 25!

Funny fact, both Jordan and James won their very first NBA title 2 years after their best season (stats wise)
23 + 2 = 25 (for many reasons)


In the following dashboard, I've tried to compare their Regular Season games setting 3 parametric bars that you can use and modify to see how many games Jordan and James finished with at leat X Points, Y Rebounds and Z Assists.
The color of that game will be lighter then the others (red for Jordan, yellow for James).

Have fun with it!

By the way, you could notice how different was their made shots distribution: Jordan barely scored 0,3 3Point field goal per game (James 1,7), probably as a consequence of his slashing style back in his prime with 11,6 2Point field goal made (James 8,4).

PS: Here's a quick Version of the dashboard for Mobile Phones:!/vizhome/MJLBJ-232isthemagicnumber/MJvsLBJ

mercoledì 19 ottobre 2016

Ten Years of Nba Draft

Scatter Plots: a brief Introduction

When you have to deal with a dataset containing lots of rows and multiple metrics, Scatter Plots may be the solution to all your Viz problems.

With a Scatter Plot chart, you can plot all your rows/detail considering a minimum of 2 metrics (X and Y coordinates) till a maximum of 4.

You may say: what could the other 2 metrics affect?
I shoud say: Size of your "plotted" shape, and eventually its color.

For example you shuold plot multiple circle (one for each row of yours) with a specific size determined by a third metric, and a certain color coming from a fourth metric (eg: gradient from worst to best).
Alternately, you could even bind colors to a dimensions.


Is it possible to evaluate how good (or bad) NBA teams have selected their picks in the last 10 years?

I tried to answer this question using 2 very well known stats: Avg. Minutes Played and Wins Share per Year of every player drafted in this span of time.

With this assumption, the more a team (first scatter plot) or a player (second scatter plot) is on the top-right corner the more they are considered in a good position with high value for Avg. minutes played and hig Wins Share per Year.
On the other end, being in the bottom-left corner means that a team has selected low profile/impact players; meanwhile for a player means that his careers is just sub-par.

You can interact with the dashboard both selecting a year to filter a specific draft class and clicking on teams logo to consider just those players drafted by that team.

Filters and Table Calculation: Tips for Tableau beginners

When you start using Tableau, dragging and dropping stuff, you see rainbows and unicorns everywhere.

It's definetly my favourite DataViz tool ever.

But after a while, when you begin to elevate your game, you may find some difficulties if you don't spend some time understanding how Tableau works, even with simple tasks.

I remember my first impasse when I had to calculate percent of total while filtering the same dimension, with a sample data like this:

Color     Value
Blue       30
Red        20
Green     40
Yellow   10

After having added Table Calculation on Value using "Percent of Total" I simply added Color on the filter shelf and selecting Green I thought I would have got 40%.
Surprisngly (for me back then, not now), I got 100%.

Why? I couldn't undesrtand how it could happen.

So I started spending some time studying Tableau and I realized of it works in terms of order of operations.

Since Percent of Total is a Table Calculation, it is executed after dimension filters (eg: selecting one of the 4 colors).
So my table calculation was working on pre-filtered data with just a color selected and 100% value was correct (40 out of 40).

To by-pass this, you just need to create a Calculated Field using a Table Calculation based on the dimension you want to filter.
With this little trick, your filter will be executed after the Percent of Total Calculus and selecting Green you'll get the 40% I was looking for back then.

If you don't trust my words (you erethic!), take a look at this simple dashboard I realized just for you.

At the end of the day, my suggestion is to study how Tableau works from the early beginning because you will save time doing you vizzes and you'll understand how powerful it could be.

martedì 18 ottobre 2016

USA Election: breaking news from the HEXagon

For this first post, I'd like to share my most recent Tableau dataViz project inspired by MakeOver Monday.

If you don't know this weekly contest, I suggest you to take a look at the link and give it a try.
Long Story Short: you have to share your personal idea starting with the same dataset coming from a previously published article.
The authors of this idea suggest to complete the work in less than 2 hours: the important thing is to communicate your own flavour, even if it's not perfect.

For this episode of MoM, the theme was USA Election.
The dataset was quite simple:
  • 2 dimensions: Date (from April 1st to October 12nd) and State
  • 4 metrics: Clinton, Trump, Other, Undecided (expressed as a % across the single row)
Since I've always been interested just in latest exit-pool, I decided to consider just the last date (October 12nd) for every state in order to have a sort What-If the election day was that date.

As you already known, USA have 52 states with completly different dimensions and sometimes filled map do not represent them in the best way because small states (eg: NY area) are always too small compared to big ones (eg: Texas).

So this time, I opted for a tiled grid map using the same size for every state which, by the way, displayed with hexagons, looks really cool for the USA (I can't say the same for Italy... already tried).

The only thing you need to have to realize this viz is a simple xlsx/csv file with the X/Y coordinate to plot every state in the right position.
I suggest you to pay attention to the shape you're going to use in order to adjust your coordinates: using hexagons, you may want to consider X-Y with decimals to create a sort of good looking beehive.

When I started this viz I wanted to show which states were a no-brainer for Clinton or Trump and find out the ones still in the middle of the process.
So basically, if the difference between Clinton and Trump %'s was greater than Undecided plus Other, a state went to the candidate with higher % (Blue or Red), otherwise the candidate with the higher % got the advantage (light Blue and light Red).

Quick and straight.

Is this the best way to visualize this dataset? Probably no, but It helped me answering my own question.
I think this should be the spirit of every viz!