Page 1 of 6

Transactions on Machine Learning and Artificial Intelligence - Vol. 10, No. 6

Publication Date: December, 25, 2022

DOI:10.14738/tmlai.106.13492. Kitagawa, K. (2022). Comparison of Machine Learning Algorithms for Ball Velocity Prediction in Baseball Pitcher using a Single

Inertial Sensor. Transactions on Machine Learning and Artificial Intelligence, 10(6). 09-14.

Services for Science and Education – United Kingdom

Comparison of Machine Learning Algorithms for Ball Velocity

Prediction in Baseball Pitcher using a Single Inertial Sensor

Kodai Kitagawa

Mechanical and Medical Engineering Course

Department of Industrial Systems Engineering

National Institute of Technology, Hachinohe College, Japan

ABSTRACT

Ball velocity of pitching is an important factor in baseball players. Commonly, ball

velocity measurement requires specific devices such as radar gun. On the other

hand, Gomaz et al. developed the accurate ball velocity measurement using two

inertial sensors on pelvis and trunk. Recently, smartphone installed inertial sensor

is popular device in daily life. Therefore, if ball velocity can be measured by only a

single inertial sensor, baseball players can measure own ball velocity by only

smartphone in daily life and various situations. Thus, the objective of this study is

to propose and evaluate the ball velocity prediction method using the only a single

inertial sensor. The proposed method predicts ball velocity using by a single inertial

sensor and machine learning technique. Five machine learning algorithms (linear

regression, support vector machine, gaussian process, artificial neural network,

and M5P) predicted ball velocity by data of single inertial sensor, body height, and

body weight. In this study, Gomaz et al.’s public data for ball velocity and inertial

data during pitching of baseball players were used for this investigation. Sensor

placement was either sternum or pelvis. Accuracy of prediction was evaluated by

root mean square error (RMSE) between actual and predicted value via leave-one- out cross-validation. The results showed that greatest algorithm (M5P) could

accurately predict ball velocity by only single inertial sensor and body parameters

(RMSE < 2.0 mph). These results suggest that ball velocity can be measured by only

single inertial sensor such as smartphone.

Keywords: Ball Velocity; Prediction; Machine Learning; Inertial Sensor.

INTRODUCTION

The ball velocity of pitching is an important factor in baseball players [1–3]. For example,

increasing pitching the ball velocity contributes to delay the batter’s reaction time [1]. In

addition, Whiteside et al.’s reported possibility that maximum ball velocity contributed to

success of pitchers [2]. Furthermore, the ball velocity was useful for rehabilitation program of

elbow in baseball pitchers [3]. From these backgrounds, to measure ball velocity during

practice is important for baseball pitcher.

Commonly, the ball velocity measurement requires specific devices such as radar gun [4]. In

addition, sensor-based instrumented ball was developed to measure ball velocity [5]. The

measurement method for ball velocity without specific devices is necessary for that various

players understand own ball velocity in daily practice. Gomaz et al. developed the

Page 2 of 6

10

Transactions on Machine Learning and Artificial Intelligence (TMLAI) Vol 10, Issue 6, December - 2022

Services for Science and Education – United Kingdom

measurement method for ball velocity using two wearable inertial sensors [4]. An inertial

sensor is installed on smartphone that is popular device in daily life, and inertial data obtained

from smartphones is was useful to measure sports activities [6–8]. Therefore, if ball velocity

can be measured by only a single inertial sensor, baseball players can measure own ball velocity

by only smartphone in daily life and various situations. Thus, the objective of this study is to

propose and evaluate the ball velocity prediction method using the only a single inertial sensor.

Ball Velocity Prediction

Figure 1 shows the ball velocity prediction method using the only a single inertial sensor. This

method predicts ball velocity by machine learning technique and inertial sensor on trunk or

pelvis. Usefulness of these sensor placements were founded by Gomaz et al.’s study [4]. In this

study, these placements are compared to reduce number of sensors.

Machine learning technique is applied for the proposed method because machine learning

technique is known as effective to understand various baseball parameters such as

sabermetrics and kinematics [9–11]. Furthermore, Gomaz et al.’s method for ball velocity

prediction used machine learning techniques too [4]. In this study, common five machine

learning algorithms (linear regression, support vector machine, gaussian process, artificial

neural network, and M5P) are compared for the proposed method using a single inertial sensor.

Euclidean norm of peak angular velocity obtained from inertial sensor was used for feature of

machine learning techniques. Body height, body weight, and age were used features too. These

features were selected based on Gomaz et al.’s dataset that was used for training data [4].

Figure 1. Overview of the Ball Velocity Prediction using a Single Inertial Sensor.