galcomis/nbaall

Link to the YouTube video :https://www.youtube.com/watch?v=SITuIj4fT1w Project Overview : This project performs an Exploratory Data Analysis (EDA) on the historical NBA All Seasons dataset in order to understand patterns that may influence Rookie Points Per Game (PPG). The analysis focuses on identifying whether the Draft Pick Number has a meaningful effect on PPG, and how different physiological and performance variables in the dataset may support or complement this understanding. The dataset includes historical draft attributes (year, pick number, round), player demographics (age, height, weight), and season-by-season performance statistics (PPG, AST, REB, and advanced metrics) for thousands of players.

Dataset Featurers The physiological and draft-related features include: Age, Gender, Player Weight, Player Height, Draft Year, Draft Round, Draft Number, College.

In addition, the dataset includes performance-related features such as: Games Played (GP) Points Per Game (PTS) Rebounds Per Game (REB) Assists Per Game (AST) Net Rating Usage Percentage True Shooting Percentage

Throughout the project, a complete data-cleaning and preprocessing workflow was carried out, which was essential for isolating the relevant rookie data. This process included removing records where a valid draft number was absent ('Undrafted'), filtering players who did not meet the minimum Games Played (GP) threshold, and strategically handling outliers in the scoring data (PPG). The central goal of the project was to examine the correlation between a player's draft and physical attributes and their immediate performance, with a specific focus on the variables: Draft Number, Points Per Game (PPG), and supporting attributes such as Height and College.

Research Question What is the strength and direction of the statistical relationship between Draft Pick Number and Points Per Game (PPG)? Its importance to the research is that it allows us to quantitatively measure the accuracy with which the NBA scouting system succeeds in predicting immediate scoring performance.

Graphs and insights from the EDA process: How is the performance variable (PPG) distributed among rookie players, and is there any skew in the data? The histogram shows a strong right-skew, indicating that the majority of rookies score low PPG (around 3-5), while a small minority achieves exceptional scores. This highlights that the average PPG is pulled up by this small, high-performing elite group.

What are the data range and extreme values (Outliers) for the PPG and Draft Number variables, and how do they reflect variability? The Box Plots illustrate the high variability in both variables. The PTS graph specifically highlights outliers (Superstar Rookies). The analytical decision to retain these outliers ensures the correlation reflects the maximum potential and true scope of the Draft Pick vs. Performance relationship.

does a player's height have a statistical relationship with their Rookie Points Per Game (PPG)? The correlation is very weak (r = 0.1). The scatter plot shows that height alone is not a strong predictor of scoring performance in the rookie season, as top scorers come from diverse heights.

Do players from the top 10 colleges (by player count) show a distinct distribution in their Draft Pick Number?Do players from the top 10 colleges (by player count) show a distinct distribution in their Draft Pick Number? Yes. The Box Plots demonstrate that elite programs like Kentucky and Duke have a consistently lower median draft number (higher draft position). This confirms that players from these programs are typically scouted and selected earlier in the draft.

Research question : What is the strength and direction of the statistical relationship between Draft Pick Number and Points Per Game? This scatter plot is the primary finding of our research. The red regression line clearly descends from left to right, demonstrating a moderate-to-strong negative correlation (r = -0.47) between our key variables. The conclusion is unambiguous: the higher a player is drafted (lower pick number), the higher their Rookie Points Per Game (PPG) will be. Despite the scatter of individual data points, the overall relationship is robust, fully confirming our original hypothesis.