Visual Analysis of NBA Player Data

Jiazhi Di, Ben Wang

, Hua Hu, Qihang Zhao and Yinggui Wang

School of Information Science and Engineering, Hangzhou Normal University, Hangzhou, Zhejiang, 311121, China

hh@hznu.edu.cn, zqh@stu.hznu.edu.cn, 2276598112@qq.com

Keywords: Visual Analytics, Professional Basketball Leagues, Sports Data, Efficiency Criteria, Radar Chart Analysis.

Abstract: Competitive sports data visualization and analysis is an important technology in sports science research.

Taking the players of the National Basketball Association (NBA) as the research object, this paper focuses on

how to visually design the comprehensive ability of a single player and multiple players. This article uses the

efficiency criterion to analyze whether a player is efficient in a period of time, so as to judge whether the

player is the best. This paper figures out the comprehensive ability of NBA players by data table, line chart,

bar chart, and radar chart, and finds out the weakness of players from the visualization chart. Furthermore,

different aspects of James and Durant are visually analyzed in detail, such as rebounds, assists, shooting rate

and distribution.

1 INTRODUCTION

As an important direction of sports discipline

research, competitive sports data analysis research is

directly related to the development of competitive

sports and the improvement of the sport discipline

system. As one of the most popular competitive

sports, basketball, especially NBA, has a great

influence in the world. All kinds of data analysis

related to NBA aims to fully excavate valuable

content from the huge, complex, chaotic and

disorderly game data information. It assists the coach

and the team to make reasonable judgments for the

team to win. With the continuous accumulation of

game data, how to improve the team’s team analysis

and decision-making ability, and transfer a large

amount of game data into rich knowledge, so as to

help the team make the most effective decision has

become the most worthy research issues in this field.

In various visual analysis of NBA, R.Maheswaran

and other scholars deconstructed the process of

rebounding (Robertson, Card, Mackinlay 1989), and

Rangel defines a method to characterize basketball

players as versatile or expert based on 13 game-

related statistics (Rangel, Ugrinowitsch & Lamas

2019). At same time, P.Maymin explained the

player’s acceleration process (Maymin 2013), and

Maheswaran utilize a heat map to represent the

proportion of offensive rebounds distributed in the

basket area, colour coded of the percentage of

offensive rebounds (Maheswaran, Chang, Henehan

2015). Meanwhile, Bashuk proposed a method to

predict the performance of the game (Bashuk 2012),

and Lorenzo classifies players' performance in the

NBA based on their anthropometric attributes and

game experience (Zhang, Lorenzo, Gómez, Mateus,

Gonçalves, Sampaio 2018).

2 EVENT STREAM DATA

VISUALIZATION

The NBA stadium data analysis method based on the

event stream decomposes the NBA game into an

ordered collection of different events, which is the

NBA event stream. Events are divided into single

player events and multi player events, that are

arranged in order according to the game time.

Through the official event data of NBA, it is found

that the data records are in chronological order. A

corresponding event at each time point is called point

event. There are 16 types of point events in Table 1.

To better describe the nature of time, each event has

its own attributes.

The point attributes include:

• Type of event: The name of the point event.

Time: The time when the point event occurred;.

Di, J., Wang, B., Hu, H., Zhao, Q. and Wang, Y.

Visual Analysis of NBA Player Data.

DOI: 10.5220/0011768800003607

In Proceedings of the 1st International Conference on Public Management, Digital Economy and Internet Technology (ICPDI 2022), pages 835-840

ISBN: 978-989-758-620-0

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

835

Table 1: POINT event types.

Start of the

ame Mistake Sus

Backboar

Two-point shot Two-point missed Three-point shot

The three-point

misse

Offensive fouls Personal fouls Technical fouls Shooting fouls

Free throw Free throw missed Sent off Timeout

Table 2: Part of The Data About Player James

season

shooting

ercenta

three-point field

oal

ercenta

free throw shooting

ercenta

rebound assist steal turnover score

2005 48% 33% 74% 8. 1 5. 8 1. 38 5. 00 30. 8

2006 42% 28% 76% 8. 1 8. 0 1. 70 3. 30 25. 1

2007 41% 26% 73% 7. 8 7. 6 1. 77 4. 15 28. 2

2008 51% 33% 75% 9. 1 7. 3 1. 64 2. 71 35. 3

2009 50% 40% 73% 9. 3 7. 6 1. 73 3. 82 29. 1

2010 47% 35% 76% 8. 4 5. 9 1. 67 3. 14 23. 7

2011 50% 26% 74% 9. 7 5. 6 1. 87 3. 52 30. 3

2012 49% 38% 78% 8. 4 6. 6 1. 78 3. 04 25. 9

2013 57% 41% 81% 7. 0 4. 8 1. 80 3. 05 27. 4

2014 42% 23% 73% 11. 3 8. 4 1. 65 4. 10 30. 1

2015 53% 34% 66% 9. 5 7. 6 2. 33 3. 57 26. 3

• Coordinates: The coordinate point at which

the event occurred. For example, player shot

charts；Normally, the stadium is a coordinate

map.

• Players: Players in point events.

• Description: A detailed description of the

point event.

The stream event attributes include:

• Type of event: Possession of the ball.

• Start time: The time when the flow event

occurred.

End time: The time when the stream event ends

3 NBA PLAYER DATA ANALYSIS

3.1 Data Acquisition

There are many websites recording sports data at

home and abroad. For instance, the Espn.com website

records the detailed data of NBA games. In addition,

Basketball-Reference.com, China.NBA.com official

website, Hupu.com websites have rich records of

relevant game data. Therefore, most of the

competition data can be obtained on the public

website, and the data in the web page can be stored

locally through a simple tool. In this paper, the

Python based HTML parsing tool with lxml library

and XPath has been implemented to grab the relevant

data from the Hupu.com website and store it locally.

3.2 Data Analysis

3.2.1 Single Player Analysis

Single-player analysis is designed to analyze a

player's performance and trends over a period of time.

It analyzes whether a player has been effective on the

court over a period of time, or whether he can help

the team score on offense and reduce the opponent's

score on defense. Certain player characteristics, such

as points, turnovers, blocked shots, assists, and

rebounds, will be focused in the experiments. The

player’s characteristics fall into two main categories:

offensive characteristics (point, turnover, assist) and

defensive characteristics (steal, block).

In time period analysis, specific characteristics

can be selected to analyze a player's performance

during that time period. In addition, the player's

efficiency rule can be used to describe its

performance in this period, and it can be displayed in

the player performance view. Efficiency rule include

points, rebounds, assists, steals, blocks, shots, field

goals, free throws, free throws made, turnovers, and

the number of games a player finished. The efficiency

rule is defined below:

efficiency rule＝[(points + rebounds + assists + steals

+ blocks) - (shots-field goals)

- (free throws-free throws made) - turnovers] / time

As shown in Table 3, James from the Lakers team

is selected and the efficiency rule technical indicator.

ICPDI 2022 - International Conference on Public Management, Digital Economy and Internet Technology

836

Table 3: James's efficiency rule trend.

time 2021-03-17 2021-03-19 2021-03-21 2021-05-01 2021-05-03

efficiency rule 1.081 1.290 1 0.656 0.714

Figure 1: James's scoring trend.

Figure 2: James' shot distribution for the 19-20 season.

(ERTI) is selected. The line chart shows his

efficiency rule chart for the last five games in 2021.

According to the chart, James' efficiency rule

fluctuated in the five major games between March

and May. At that period, the Lakers have five new

players in the rotation this season, and the playoffs are

the shortest in history, leaving little time for players

to fit in and rest. Figure 1 shows James' average points

per game in the playoffs each year. From the official

website, the NBA league average is 10.4 points per

game. As shown in the chart, James' playoff scoring

per game per year is much higher than the league

average. In 2008, he scored his best points per game,

35.3 points. Other years fluctuated around 29.04

points. and the range of fluctuating tends to be stable,

which shows James has a high scoring ability

Figure 2 shows the distribution of James' shooting

statistics for the 19-20 season. James' shooting

percentage of 49.8% for the 2019-2020 season

remains normal. Also shooting 34.9% from 3-point

range is normal. The chart shows that James is hitting

more shots from long range than from mid range, and

confirms that scoring further and further away from

the basket is the way James is scoring now.

3.2.2 Multi-player Analysis

Basketball is a team sport, which requires

consideration of the collective performance of

multiple players. At the same time, coaches and NBA

analysts can select multiple players and skills to

compare different players for visual analysis. The

segmentation and selection of event segment in single

player analysis is further extended in multi-player

visual comparative analysis. Furthermore, coaches

and analysts can select a time period of interest to

Visual Analysis of NBA Player Data

837

Figure 3: James hitting vs Durant hitting.

look at individual versus multiple skill

comparisons, and see how different players

performed in a particular skill during that time period.

Figures 3 shows a comparison between James and

Durant in three aspects: shooting percentage, three-

point field goal percentage, and free throw shooting

percentage. As James was in the league four seasons

before Durant, he was a first-round pick in 2003, and

Durant was a top pick in 2007. So in terms of time,

for a better comparison we have chosen the years

2009-2017 when James and Durant shared stats. In

terms of free throws, Durant's shooting percentage in

the eight years has basically stabilized at more than

80%, but James' free throws may not be his strong

point. Especially in 2016, James only got 66%. in

terms of shooting. At the same time, Durant is on a

steady upward trend, with the highest shooting

percentage of 55.6% in 2016.

Rebounding refers to the ball that rebounds from

the backboard or hoop after a miss. In fact,

rebounding is a more complicated technique, which

is composed of positions, take-offs, air grabs and

actions after the ball is obtained. Rebounding is an

important part of offensive and defensive tactics in

basketball games. Figure 4 is a histogram of the

changes in rebounds between James and Durant from

2009 to 2018. The figure shows clearly that James is

much better than Durant in rebounding. James has a

stable average rebounding score of 9.09 in the past

nine years, while Durant has only 7.66. Normally,

rebounding is an important way to control the ball and

it has a direct impact on the outcome of the game.

James can get more rebounds means that his team can

get more ball rights, so that the team has a greater

winning percentage.

Figure 5 shows the data of time, games, scoring,

fouls, turnovers, blocks, steals, assists, rebounds, free

throw percentage, three-point shooting percentage,

and field goal percentage since James and Durant

played in the NBA. Figure 5 is a radar chart drawn by

radar analysis method, which is an effective method

for systematic analysis of business operations. This

method analyzes the company's operating results

from five aspects: its operating profitability, safety,

liquidity, productivity, and growth. The relevant data

of these five aspects are expressed by ratios and filled

in an equal scale graph which can show the

relationship of their ratios. After connecting the nodes

of their ratios with colored pens, it is like a radar chart.

From the figure, we can see the whole picture of the

player’s skills, directly find out his weak links, and

lay the foundation for the next step of improvement

(Zhang, Jia 1990). Through the comparison, Durant's

weak links are mainly assists, steals and rebounds,

and James’s weak links are free throw shooting

percentage, cap, and foul.

Actually, James' defense is very strong. So far in

his career, James has been selected to the all-

defensive team for 5 times. However, Durant has a

very physical advantage. He has the height and

wingspan of a center and the speed of a defender.

Offensive players have thought that it is not a simple

matter to break through Durant. Generally speaking,

Durant is relatively thin and a bit worse in defense. In

terms of leadership ability, James is better than

Durant. James is able to connect the entire team, and

the team are arranged around him. Durant's ability to

lead the team is relatively insufficient. In Thunder

team, he was talented but failed to win the

championship. Durant's job in the Warriors was just

scoring, not to lead the team.

ICPDI 2022 - International Conference on Public Management, Digital Economy and Internet Technology

838

Figure 4: James rebound vs Durant rebound.

Figure 5: Comprehensive comparison.

4 CONCLUSIONS

Competitive sports data analysis is an important part

of sports research. In this paper, NBA data

visualization uses figures to represent abstract

competitive data. The application of visual effects

helps people to enhance the cognition and

understanding of abstract information. With the

development of visualization technology, the visual

system will be more effective and efficient. The

visualization system for professional analysts can not

only realize the real-time reproduction of the game

process, but also automatically analyze and present

the time period of interest to professional users. The

analysis of single player, multiple players and the

cooperation among players can find the strengths and

weaknesses of players. In addition, the system not

only helps to analyze the opponent's strategy, but also

can predict the internal instability of the team. Visual

analysis system will become an indispensable tool in

the field of professional sports.

ACKNOWLEDGMENTS

This paper was funded by projects: Zhejiang Province

(lgf19f020011, Y202044936); Hangzhou

(20191203b14). And it utilized the research platform

and data in our teams (Chen, Wang, Wang, Wang

2021, Wang, Wang. Huang 2020).

Visual Analysis of NBA Player Data

839

REFERENCES

Bashuk, M. (2012). Using cumulative win probabilities to

predict NCAA basketball performance. In: Proceedings

of the MIT Sloan Sports Analytics Conference. Boston.

pp. 1-10.

Chen, L.Q., Wang, B., Wang, Y.G., Wang, X.Y. (2021).

Exploratory Data Analysis on the Usage of COVID-19

Vaccine. In: Proceedings of the 6th ISCIPT Symposium

on Computer and Information Processing Technology.

Changsha. pp. 101-105.

Maymin, P. (2013). Acceleration in the NBA: Towards an

algorithmic taxonomy of basketball plays. In: MIT

Sloan Sports Analytics Conference. Boston.

Maheswaran R, Chang, Y.H., Henehan, A. (2015)

Deconstructing the rebound with optical tracking data.

In: MIT Sloan Sports Analytics Conference. Boston.

Robertson, G., Card, S.K., Mackinlay, J. D. (1989). The

cognitive coprocessor architecture for interactive user

interfaces. In: Proceedings of the 2nd annual ACM

SIGGRAPH symposium on User interface software

and technology. Boston. pp. 10-18.

Rangel, W., Ugrinowitsch, C., & Lamas, L. (2019).

Basketball players' versatility: Assessing the diversity

of tactical roles. International Journal of Sports Science

& Coaching. 14(4): 552-561.

Wang Y.G., Wang. B., Huang, Y.X. (2020). Comprehensive

analysis and mining big data of smart ecommerce user

behavior. Journal of Physics, 1616(1):1-7.

Zhang, S., Lorenzo, A., Gómez, M.A., Mateus, N.,

Gonçalves, B., Sampaio, J. (2018). Clustering

performances in the NBA according to players’

anthropometric attributes and playing

experience. Journal of sports sciences. 36(22): 2511-

2520.

Zhang, H.J., Jia, C.M. (1990) The Chinese Dictionary of

Auditing. LiaoNing People's Publishing House, The

ShenYang.

ICPDI 2022 - International Conference on Public Management, Digital Economy and Internet Technology

840