This project leverages Machine Learning and Statistical Analysis to predict NBA player performance (Points Scored) based on the 2024-2025 season statistics. By analyzing metrics such as usage rates, shooting efficiency, and playtime, the model provides accurate scoring forecasts using Linear Regression algorithms.
The primary objective is to build a predictive model that estimates a player's Points (PTS) output. The project follows a rigorous Data Science lifecycle:
- Data Collection: Gathering 2024-2025 active player stats.
- Exploratory Data Analysis (EDA): Understanding feature distributions and correlations.
- Statistical Validation: Testing for normality (Kurtosis/Skewness) to ensure model reliability.
- Modeling: Training Linear Regression and OLS models.
To determine which factors most heavily influence scoring, I generated a correlation matrix of all features.
Insight: The heatmap confirms that Field Goals (FG), Field Goal Attempts (FGA), and Minutes Played (MP) have the highest positive correlation with Points (PTS).
Before applying Linear Regression, I analyzed the distribution of key variables. The histogram below shows the distribution of Field Goals (FG). The curve helps visualize the skewness and kurtosis, guiding the decision on feature scaling and normalization.
##Top Performers Analysis
As part of the analysis, the system identified and ranked the top performing players of the current season based on a composite score of their stats (Game Score, Points, Assists, Rebounds).
Observation: Nikola Jokić, Shai Gilgeous-Alexander, and Giannis Antetokounmpo lead the league in efficiency and scoring impact.
The Linear Regression model was tested on a 20% holdout dataset. The scatter plot below demonstrates the relationship between the Actual Points (X-axis) and the Predicted Points (Y-axis).
- R² Score: ~0.95 (High Accuracy)
- Trend: The points tightly follow the red diagonal line, indicating that the model generalizes well and makes highly accurate predictions with minimal error variance.
- Language: Python
- Data Manipulation: Pandas, NumPy
- Visualization: Matplotlib, Seaborn
- Machine Learning: Scikit-Learn
- Statistics: SciPy, Statsmodels (OLS Regression)
-
Clone the repository
git clone [https://github.com/yucezis/NBA-Score-Prediction.git](https://github.com/yucezis/NBA-Score-Prediction.git) cd NBA-Score-Prediction -
Install dependencies
pip install -r requirements.txt
-
Run the Jupyter Notebook
jupyter notebook NBA_Score_Prediction.ipynb
- Name: Zişan Yüce
- Role: Computer Engineering Student
⭐ If you like this project, don't forget to give it a star!