Combining Radar and Optical Sensor Data to Measure Player Value in Baseball.

Area of Science:

• Sports Analytics
• Baseball Performance Metrics
• Machine Learning in Sports

Background:

• Evaluating baseball player talent from batted ball data is challenging.
• Major League Baseball stadiums collect vast amounts of game data via sensors.
• Existing models like weighted on-base average cube use batted ball parameters, but running speed's impact is recognized.

Purpose of the Study:

• To develop an improved model for quantifying baseball player batted ball performance.
• To integrate multi-sensor data, including running speed, into performance evaluation.
• To leverage machine learning for enhanced predictive accuracy in baseball analytics.

Main Methods:

• Utilized machine learning techniques to combine diverse data streams.
• Integrated three-dimensional batted ball vectors from Doppler radar.
• Incorporated running speed measurements from stereoscopic optical sensors.

Main Results:

• The combined data model demonstrated improved accuracy in evaluating batted ball performance.
• Machine learning effectively synthesized information from radar and optical sensors.
• Running speed was shown to be a significant factor in batted ball outcomes.

Conclusions:

• Combining batted ball dynamics with running speed offers a more comprehensive player evaluation.
• Machine learning provides a powerful tool for integrating complex sensor data in sports analytics.
• This approach enhances the ability to quantify and predict player performance in baseball.