Sport Performance Analysis 10/7

Unplanned Discussion and Data Analysis in Sports

Introduction

  • The speaker reflects on a spontaneous topic arising during a discussion.

    • Mentioned attempting to think of a specific subject related to basketball but unable to recall it immediately.

Chapter 36: Expected Point Value of Plays

  • Discussed the expected point value of basketball plays:

    • Comparison of pick and rolls versus post-up plays.

    • Mention of Shaquille O'Neal and the five-second rule changing gameplay.

    • Statistics indicate shots taken from outside the three-point arc yield higher average points per shot compared to post-up shots.

  • Mention of a recent article suggesting that the data analysis around taking three-pointers might be misleading:

    • Key points about getting fouled more often during two-point shots, leading to potential free throws that could affect scoring expectations.

    • Acknowledgment of evolving strategies and potential shifts in gameplay dynamics may occur.

Chapter 38: Trailing by Two Points

  • Discussion of strategies when trailing by two points in games:

    • Whether to go for a tie (two-point shot) or win (three-point shot).

    • Consideration of win probabilities in overtime if the game is tied at the end of regulation.

  • Examples from football where away teams may be encouraged to pursue a two-point conversion to increase their winning chances.

  • Discussion included considering analytics tools from ESPN to facilitate decision-making based on historical data.

  • Contains reflection on the psychological aspects of risk in decision-making at critical game moments.

Discussion on Free Throws and Post-Ups

  • Analysis of players taking fouls when shooting, particularly post-up players:

    • Historic data showing big players underperforming in free throw shooting, impacting the value of post-up play.

    • Reference to players like Joel Embiid versus past players like Shaquille O'Neal regarding their performance and free throw capabilities.

  • Arguments about current strategies employed by teams versus historical patterns, acknowledging that player skill sets and stats evolve over time.

Data Preparation for Statistical Analysis

  • Introduction of a new book on sports analytics, focusing on implementing data using R:

    • The instructor contemplates using this new book for future classes.

  • Tasks involve downloading files for exercises and learning to employ R through dataset analyses in basketball.

  • Discussion moved toward practical hands-on activities with data, including:

    • Loading R data files, manipulating datasets, calculating metrics like effective field goal percentage, and understanding Dean Oliver's four factors of basketball success:

    • Effective field goal percentage.

    • Turnover percentage.

    • Offensive rebounding percentage.

    • Free throw rate (not percentage).

  • Calculation steps including interpreting how to break down total field goals into two-point and three-point shots for effective calculations.

Merging Datasets in R

  • Instruction on merging multiple datasets:

    • Example of using left join function to aggregate game statistics with advanced team stats.

    • Insight into the resulting columns and potential data manipulation resulting from merging.

  • Discussion about handling non-matching entries in merged dataframes and common pitfalls.

  • Provided definitions for various joining functions (left join, right join, inner join, outer join) in R, with implications for data analysis regarding NBA stats.

Visualization Techniques in Data Analysis

  • Conducting scatter plots, density plots, and creating comprehensive visualizations of statistical relationships:

    • Discussed how comparisons across metrics would yield insights into player performance, strategies in games, and overall team efficiency.

  • Mentioned utilizing ggplot functions to visualize effective field goal percentages and how different teams compare.

  • Notably addressed creating faceted plots to show distributions across multiple variables simultaneously to derive conclusions in player performance.

Clustering and Model Interpretation

  • Description of hierarchical clustering methods applied to draft data:

    • Aiming to observe player performance categories followed by dendrogram outputs showing relationships between picks.

  • Discussion about how clustering informs team strategies in selection and training - highlighting differences observed per player and statistical changes over time.

  • Examples of statistical regression modeling for win prediction outcomes in basketball.

Closing Remarks on Analytical Techniques

  • Introduction to different basketball project ideas students can explore on Kaggle, promoting independent work and research.

  • Examples provided by students demonstrated their engagements with various basketball datasets and models:

    • Notably emphasized on player evaluations, performance statistics, and trend analysis across seasons and player selections.