Table of Contents

• Data Scraping Script
  • Notes about the script
  • Variables Notes
• LeBron Analysis
  • Graphs and Takeways
  • Conclusions
• Author's Thoughts
• Other Libraries Used
• Authors
  • Acknowledgements

NBA Data Analyzation

A short project that examines player statistics collected from basketball reference.

Data Scraping Script

There is a python script that scrapes data and takes in a user-inputted range, ranging from 1980-2019. Prior to 1980, stat-tracking was spotty and some statistics were completely ignored altogether. For example, steals and blocks were not recorded until the 1973-74 NBA season.

The script makes sure that the user doesn't enter an incorrect year, and compiles the data from basketballreference from any of the following pages:

• Advanced Stats
• Per-36 Minute Stats
• Per-100 Possession Stats
• Per-Game Stats
• Season Totals

Notes about the Scraping Script

• Files will be outputted in JSON format, with the naming schema STARTyear-ENDyear_stat_playoff/regularSeason.JSON.

• Due to the nature of the actual source of the data, I encountered the problem of some of the fields in the HTML table (namely fields that involved ratios) outputting empty strings because of an empty field in the HTML itself.

  • In order to combat this, and to make sure that the entirety of a row was taken if it contained data, code in the scripts actually scraping the data looked like the following:

if row.select_one('[data-stat=fg3_pct]').text.strip() == "":
          d['3P_PCT'] = float(0)
      else:
          d['3P_PCT'] = float(row.select_one(
              '[data-stat=fg3_pct]').text.strip())

• This isn't necessarily the most efficient way to get around the problem but it was the simplest one I could think of at the time. For this reason, data collection may be more resource intensive than I initially desired.

• This may or may not be optimized in the future, but as of writing this I opted to leave this alone as the script works fine.

• Increasing the range of the number of years you're scraping will significantly impact the amount of time you're waiting for the script to do its job.

• There are different functions for each type of page, but these can only be called by the main function, bbref_scrape. It's not suggested to actually use these functions individually. The intention was to just automate this for anyone who wants to get data easy.

• The actual script can just be called using the function bbref_scrape, and this can be called to follow a set of prompts that fill in important variables for the other functions.

Variable Notes

There are certain variables that are repeated throughout the script. Here are their descriptions:

• reg_play - This is the variable that tells the script whether or not the data its looking for is in the regular season or playoffs.

• user_start_year/start_year - This variable is what holds the user inputted start year. It's used to fill in the year field for each data entry, create a list of pages along with the end year, and to help create the file name. The name was shortened when called by the scraping functions in order to distinguish them.

• user_end_year/end_year - This variable is what holds the user inputted end year. It's used to help create a page list and to help create the file name. The name was shortened when called by the scraping functions in order to distinguish them.

• type/tp - This variable helps tell the page list creating function what type of page list to create and it's called by that function from the individual page scraping functions.

LeBron Analysis

This analysis provides a general overview of some of LeBron's advanced statistics (mainly WS) compared to the rest of the league.

Notes About the Data

There are some general notes about the data that should be mentioned.

• The observations only include the years that LeBron made the playoffs, which were 2006-2018. This is because of the fact that I wanted to be able to easily compare regular season stats and playoff stats better.

• The data containing the rest of the league was trimmed in order to be able to better observe the higher performers in the league. I did this through the Win Share metric. Why? This metric is probably the best one as it incorporates both defense and offense in its calculations.

  • Because of this the data was filtered out. This was done by only including the players that had more than 1 WS in the playoffs, as adding a win in the playoffs is something that a lot of teams would appreciate.
  • The regular season WS was filtered differently. The 2019 season saw the Sacramento Kings miss the playoffs in the Western Conference by 10 wins and the Hornets by 2 wins in the Eastern Conference. I opted to just use the average by these two and go for anyone who contributed by 6 or more wins in the regular season.

Graphs and Takeaways

Playoff WS Comparison:

This graph shows us that compared to the rest of the league, LeBron is certainly above average in terms of WS. This doesn't tell us much besides LeBron James is a really good basketball player (duh). The one insightful thing we can see is 2012 is clearly an outlier, even by LeBron's standards. The two possible reasons for this:

1. There was a lockout that caused the 2011-2012 season to be shortened to 62 games. The fewer games meant more rest for LeBron and thus he performed better come playoff time.
2. LeBron had lost the 2011 finals in a disappointing fashion. He might've channeled this disappointment into one of his highest performing playoff performances in his career in regards to WS.

LeBron's True Shooting % compared to Usage Percentage:

This graph highlights LeBron's TS% to his USG%. Some takeaways from this graph:

• Most of the time, it seems to be that his USG% in the regular season is consistently around 30%. This makes sense as LeBron was always the best player on whatever team he was on throughout these years and he's a ballhandler.
• His playoff USG% is a different story. It's erratic to say the least but there are 2 clear outliers.
  • 2011 is his lowest playoff USG%, this is also considered LeBron's worst playoff run.
  • 2015 is his highest playoff USG%, this can be easily attributed to the fact that both of his All-Star teammates went down during the playoffs.
• 2015, coincidentally, is his worst playoff run TS%. My hypothesis? Since LeBron was the main focal point of the offense without Irving in the finals and Love for most of the playoffs, he was probably forcing shots that just weren't there and his TS% suffered as a result.
  • Generally, LeBron's TS% was above 55% in the regular season, which is very good. In the playoffs this was more erratic, but generally speaking he never dipped below 50% other than 2015.
  • LeBron's 2013-14 playoff TS% was ridiculously high. Probably as a result of the earlier rounds prior to the finals. It might be interesting to just look at his stats in his games in the finals.

WS Comparison to the best of the rest of the league:

• LeBron's WS for the regular season was the highest in the league from 2009-2013
• LeBron's WS for the most part is above 12.5, meaning that he alone added 12.5 or more total wins for the season. To put that into perspective, almost 15% or more of the wins a team that has LeBron are attributed to his being there.
  • Note that this of course doesn't discriminate between which games were actually attributed to James but instead is insightful into seeing how many wins his teams achieved because they had James
• LeBron suffered a major injury in the 2014-2015 NBA season that caused him to miss 2 weeks. This explains his lowest WS overall.
• Playoffs is much more erratic in general but in general, there were only 4 times throughout LeBron's playoff career that someone has obtained a higher WS than him. This speaks to the value of LeBron come playoff time.
• Finally, WS seems to heavily lean on players that are offensively focused. This can be seen in the regular season, as the players who had a higher regular season WS than LeBron are, generally speaking, offensive superstars. This metric isn't perfect (like all metrics) and thus it's a good idea to also look into something like Defensive Win Shares.

DWS Comparison to the best of the rest of the league:

• Generally, LeBron is not the best defender even though he is above average. The only times he had the highest DWS were in 2007, 2011-13, and 2016.
• He tied Kevin Garnett in 2012 and Draymond Green in 2016 for having the highest playoff DWS.
• Even though this indicates that LeBron isn't as superb on defense as he is on offense, defensive stats should always be taken with a grain of salt. Defense in the NBA is incredibly difficult and has a multitude of factors no statistic or metric can accurately capture. Despite this, it may be helpful to look into these statistics as a way of trying to examine player profiles purely on the surface level.

Conclusions

LeBron has been dominant during this 12 year period (2006-2018), and has been consistently the best player or one of the best players in the NBA. This is amazing as he's essentially been playing at a high level for almost the entirety of his career in the NBA. Even though he is probably not going to be remembered for his defensive prowess, he is no slouch either. And, his USG% reflects how much he actually imposes his will on whatever team he's on: if LeBron is on your roster, it's LeBron's team from then on.

Author's Thoughts

In general, this project is a good stepping stone to observing other statistics of pretty much anyone in the NBA thanks to the versatility of the script. I will most likely use what I have created and learned here for other projects in the future. I hope the visualizations might help anyone who wanted to learn how to do it just like I did at the start of this project.

Other Libraries Used

The libraries that were used for the creation of the script include:

• json
• urllib3
• choice
• requests
• BeautifulSoup
• time

The other libraries used for the analysis are as followed:

• pandas
• matplotlib

Authors

The script, data visualizations, and the data analysis write-ups were all created by Mir Khan. Find my Github page here.

Acknowledgements

I learned how to scrape data from LearnDataSci's tutorial here, check it out here.