nirm2002/BasketBallData

🎥 Click here to watch the presentation video

🏀 Predicting Basketball Game Outcomes

📊 Dataset Overview

Dataset: KDAM1/BasketballGames (From Hugging Face) Goal: To analyze basketball game statistics and determine if we can predict the outcome of a game (Win or Loss) based on a team's performance across various statistical parameters. Target Variable: won (Game Outcome)

🧹 Data Cleaning & Decisions

To ensure accurate analysis and modeling, the dataset underwent a thorough cleaning process:

• Duplicates & Missing Values: Removed duplicate entries and imputed missing numerical values using the median to avoid skewing the data. Completely empty columns were dropped.
• Standardization: Converted string columns to lowercase and stripped extra spaces for consistency.
• Normalization: Applied MinMaxScaler to scale numerical columns between 0 and 1. This is crucial for comparing different metrics (e.g., points vs. turnovers) on the same visual scale.

🔍 Exploratory Data Analysis (EDA) - Questions & Visualizations

Question 1: Do rebounds significantly impact the game outcome?

Answer: The boxplot compares the distribution of defensive rebounds (drb) between wins and losses. We can observe the median and the spread of rebounds to see if winning teams consistently secure more boards.

Question 2: How does the balance between teamwork (Assists) and mistakes (Turnovers) affect the chances of winning?

Answer: The scatter plot maps Assists (AST) against Turnovers (TOV), with points colored by the game outcome (won). This visualization helps identify if clusters of wins appear in the high-assist/low-turnover regions.

Question 3: Is there a distinct difference in Field Goals Made (FGM) between winning and losing teams?

Answer: The overlapping KDE (density) plot illustrates the distribution of Field Goals Made. By looking at the shift between the two "peaks," we can clearly see how scoring efficiency directly separates the winning teams from the losing ones.

Question 4: Do winning teams rely more heavily on 3-point shooting?

Answer: The violin plot shows the distribution and concentration of 3-Point Field Goals Made (FG3M). The shape and thickness of the "violins" reveal whether winning teams have a higher volume or more consistent 3-point shooting performance.

💡 Reflections and Insights

Transitioning the research question to focus on predicting the target variable (won) provided a much clearer direction for the EDA. Handling missing values and ensuring the data was properly normalized were critical steps that allowed for fair comparisons across different statistical categories. The visual analysis highlights that efficiency (Field Goals Made) and minimizing mistakes relative to teamwork (Assists vs. Turnovers) are strong indicators of a winning outcome. Overcoming technical hurdles, such as resolving KeyError issues after dropping columns and adapting to library updates (like Seaborn's hue requirements), were valuable learning experiences in the data wrangling process.