mlb-ops-predictions

Predicting MLB player OPS using multiple linear regression and 12 batting metrics. Includes data analysis, visualizations, and evaluation metrics to assess model accuracy and explore player potential.

Predicting MLB Batting Performance with OPS

Project Overview

This project analyzes the relationship between 12 MLB batting metrics and a batter's OPS (on-base plus slugging). Using multiple linear regression, it aims to predict OPS based on secondary metrics, offering insights for scouts, teams, fans, and players.

Motivation

OPS is a widely used metric for evaluating overall batting performance. By predicting OPS using other statistics, this project helps identify player potential, optimize team lineups, and inform decision-making in scouting and player development.

Data

The dataset was sourced from Baseball Savant, including:

• Years: 2019, 2021-2024 (excluding 2020 due to the shortened season)
• Minimum plate appearances: 502
• Metrics: Singles, doubles, triples, home runs, strikeouts, walks, batting average, runs batted in, total bases, stolen bases, barrel percentage, and in-zone percentage

Ethical Considerations

The data is publicly available and devoid of personal information, ensuring no ethical or privacy concerns.

Methodology

• Model: Multiple Linear Regression
• Evaluation Metrics:
  • Mean Squared Error (MSE): Measures the accuracy of predictions.
  • R-squared: Evaluates the variance explained by the model.

Results

• Mean Squared Error: 0.00008457, indicating high accuracy.
• R-squared: 0.9900, showing that 99% of OPS variance is explained by the model.

Visualizations include:

• Scatter plot comparing predicted and actual OPS values.
• Pair plot showing correlations between independent variables and OPS.

Key Findings

The model successfully predicts OPS with high accuracy, demonstrating the strong relationship between the chosen metrics and a player's offensive performance.

Future Work

• Pitching Analysis: Develop a similar model to predict pitcher performance using metrics like ERA.
• Expanded Metrics: Explore additional statistics to enhance the current model.

Repository Contents

• Data: CSV files sourced from Baseball Savant.
• Code: Jupyter Notebook for data preprocessing, model training, and evaluation.
• Visualizations: Scatter plots and pair plots showcasing results.

Getting Started

1. Clone the repository:

   git clone <repo-url>

2. Install the required Python libraries:

   pip install -r requirements.txt

3. Open the Jupyter Notebook to run the analysis:

   jupyter notebook mlb_ops_prediction.ipynb

Acknowledgments

• Data sourced from Baseball Savant.
• This project was created as part of a data science course at Northeastern University.