• For this project, I will be using the “cbb.csv” dataset from Kaggle. The .csv file can be downloaded here: https://www.kaggle.com/datasets/andrewsundberg/college-basketball-dataset

• The dataset above contains various (24 columns) statistics for NCAA Division 1 Men’s Basketball Teams from 2013 - 2024, excluding the 2020 Covid Season.

• We will be using the dataset above to find various trends and investigate some old sayings in basketball. Without further adieu, Let’s get started.

• If you have ever played organized basketball before, “Defense Wins Championships” is definitely a phrase you have heard from your coaches before.

• Does the data back that up? Let’s find out with a boxplot of team’s Adjusted Defensive Efficiency (ADJDE) and Postseason Success.

Champions: min: 84.5, q1: 90.675, median: 91.3, q3: 93.075, max: 94.5

2ND place: min: 85.2, q1: 89.575, median: 93.5, q3: 94.45, max: 96.2

Final 4: min: 84, q1: 91, median: 93.95, q3: 96, max: 102

Elite 8: min: 85.7, q1: 90.55, median: 93, q3: 96.6, max: 103.3

Round of 16: min: 85.4, q1: 90.6, median: 93.8, q3: 96.5, max: 107.1

Round of 32: min: 84.1, q1: 92.75, median: 95.2, q3: 98.1, max: 114.7

Round of 64: min: 84.3, q1: 95.2, median: 98.2, q3: 102.1, max: 114.5

Teams that did not qualify: min: 88, q1: 101.5, median: 105.5, q3: 109.5, max: 120.7

• As seen by the boxplot and the 5 number summary, teams that make it further in March Madness are generally better on defense, most notably the 2013 Louisville Cardinals led by coach Rick Pitino with an outlier ADJDE of 84.5.

• With that being said, there are still some teams with good defense that ended up not making very far in the tournament, such as the 2015 Virginia Cavaliers with an ADJDE of 84.1 that lost in the round of 32.

• Conclusion: the saying “Defense Wins Championships” is generally true!

• “The best defense is good offense” is definitely another phrase that you have heard before if you have played basketball before, but does good offense actually lead to good defense?

• Let’s find out with a scatter plot of team’s Adjusted Offensive Efficiency (ADJOE) and Adjusted Defensive Efficiency (ADJDE).

• Looking at the line plot above, you can somewhat see an inverse trend between ADJOE and ADJDE. Looking at the line plot summary wth a p-value of 2*10^-16 for ADJOE, indicates that the linear regression line of -0.51x+ 156.4 of the ADJOE vs. ADJDE is statistically significant and that ADJOE is a statistically significant predictor of ADJDE.

• From the linear regression line of -0.51x+ 156.4 of the ADJOE vs. ADJDE, we know that as ADJOE increases, ADJDE decreases, confirming the old saying that “The Best Defense is Good Offense”!

• To emphasize the importance of rebounding, coaches often reiterate to the players that “rebounding wins games” throughout the game.

• Just how much does rebounding impact the chances of winning a game? Does offensive or defensive rebounds matter more? Let’s find out with a 3D plot of ORB (offensive rebound rate), DRB (offensive rebound rate allowed), and win percentage.

plot_ly(data = cbb, 
        x = ~ORB, y = ~DRB, z = ~win_percentage, type = "scatter3d", 
        mode = "markers",  color = ~CONF, 
        text = ~paste("Team: ", TEAM, "</br>Year: ", YEAR), 
        hovertemplate = paste("<b>%{text}</b><br>","ORB: %{x}", 
                              "<br> DRB %{y}", 
                              "<br> Win Percentage: %{z}")) %>%
  layout(title = "ORB and DRB's Impact on a Team's Win Percentage", 
         scene = list(
    xaxis = list(title = "Offensive Rebounding Rate"), 
    yaxis = list(title = "Defensive Rebounding Rate"), 
    zaxis = list(title = "Win Percentage")), 
    width = 600, height = 400)

# Modified the hover template to show the team, the year, ORB, DRB, 
# and win percentage of each dot on the scatter plot 

• Model Coefficients:

##              Estimate Std. Error   t value     Pr(>|t|)
## (Intercept) 66.456578 4.40659008  15.08118 3.370207e-48
## ORB          1.633779 0.08867843  18.42364 5.542303e-69
## DRB         -1.906622 0.12952409 -14.72021 3.906443e-46

## [1] "R squared value of the overall regession model is:  0.24"

• Individual Correlations:

  • Correlation between ORB and win percentage: 0.3769986

  • Correlation between DRB and win percentage: -0.2914118

  • Correlation between ORB and DRB: 0.0680083

• ORB and DRB are statistically significant indicators of win percentage given that the P-values of ORB and DRB from the model coefficients are 5.54*10^-69 and 3.91*10^-46 respectively.

• The R squared value of 0.24 indicates that around 24% of the variability in win percentage can be explained by ORB and DRB.

• The larger (in terms of magnitude) DRB coefficient of -1.9 compared to ORB coefficient of 1.6 indicates that DRB has more of an impact on win percentage compared to ORB.

• Surprisingly ORB and DRB has almost no correlation with a correlation coefficient of 0.07

• Conclusion: “Rebounding wins games” is true!

• There has always been a debate on which conference is the best conference in NCAA Mens Basketball. Let’s investigate using the data we have by finding out:

1. The win percentage of each conference from 2013 - 2024, exlcuding 2020

2. Which conference has the most championships from 2013 - 2024, exlcuding 2020

• Big 12 leads all conferences in win percentage with 66.8741355%

• However, Big East had the most championships with 5, more than double of Big 12 who has only 2.

• So which conference is the best conference? Do you think overall conference win percentage or the number of championships is more important? I will leave that up to you guys to decide.

Hope you guys enjoyed some analysis on NCAA Divison 1 Men’s Basketball over the last few seasons. Turns out, scoring baskets, playing good defense, and rebounding the ball well contributes to winning basketball (Wow! What a Surprise!). In case you missed it, the source for the .csv file is on the second slide, go take a look at it sometime and maybe you will find some interesting trends too!

Thank you for your time.

https://www.kaggle.com/datasets/andrewsundberg/college-basketball-dataset

https://en.wikipedia.org/wiki/2012%E2%80%9313_Louisville_Cardinals_men%27s_basketball_team