.png)
11 hours ago
1
Visually analyze the NBA teams statistics using parallel coordinates plot. For example, analyze the impact of 2pt vs 3pt shot frequency on the number of wins in the season.
The interactive visualization is available at https://marcindulak.github.io/nba-teams-parallel-coordinates. You can open the link in a browser, no login is required.
The code uses plotly.js, and is confined to index.html file.
The screencast below uses a feature of parallel coordinates called brushing. Brushing highlights a group of lines while fading out all the others.
Various observations can be made with the help of brushing.
-
In the mid 90s to mid 00s teams were making less 3-pointers (3PM), compared to mid 10s and 20s. The average number of 3-pointers made per game increased from about 10 to 17, respectively. The number of points per game (PTS) also increased, by roughly 10 points.
-
In the recent years, the association between making more 3-pointers and winning more games (W) is more pronounced.
Another interactive feature of parallel coordinates plots, shown on the screencast, is the ability to rearrange the axis.
- Moving the FTM (free-throws made) axis between W and PTS results in a number of crossing lines. On a parallel coordinates plot, this is an indication of a negative correlation - some teams that make many free-throws (they get fouled often) score less points and win less games.
The animated gif, as the one above, can be created from a screen recording, using the command below.
For more examples of analyzing data with parallel coordinates see the Kai Chang - Visually Exploring Multidimensional Data - 2013 and Parallel Coordinates: Visual Multidimensional Geometry and its Applications - 2016 - Alfred Inselberg talks.
Note that a copy of the data is already included in this repository. The instructions below are needed to perform a data update, to add a new season statistics.
https://www.nba.com/termsofuse#nba-statistics on 2024-12-30 allow the use of statistics
for legitimate news reporting or private, non-commercial purposes
The process of obtaining statistics in the csvformat needed for visualization is described below, using the example of 2023-24 season.
Obtain the statistics of all teams in the given season. Go to https://www.nba.com/stats/teams/traditional?Season=2023-24, and mouse select the table, including the Season: 2023-24 text on the top left. Save the mouse selection as nba.com/2023-24-Traditional.txt text file, relative to the root of this repository clone. You may need to edit manually the file, so the first two lines read:
Execute the following command, to convert the nba.com/2023-24-Traditional.txt file into csv, and save as nba.com/2023-24-Traditional.csv.
If desired, analogously obtain the statistics of the opponents from https://www.nba.com/stats/teams/opponent?Season=2023-24, the txt2csv.js scripts automatically detects the Traditional and Opponents statistics file format.
Then join the selected csv files into one nba.com.csv.
