| Classification of Ptching and Arm Motion Using Data Mining |
|
In-Sub Jeong, Ki-Kwang Lee, Min-Ho Choi |
Kookmin University
Zikto |
Correspondence:
In-Sub Jeong,
Email: unginsub03@naver.com
|
|
Received: 4 January 2017 • Accepted: 29 April 2017 |
|
|
| Abstract |
PURPOSE
The purpose of this research is to classify pitching motions using the data mining method, which aims to help injury prevention overuse.
METHODS
One healthy person participated in the experiment. Subject performed six actions like pitching including pitching by wearing a smart band with IMU sensor built in the wrist. We converted the IMU data of each of the six motion into 5 Datasets. We performed data mining using the WEKA program to find the Dataset with the highest classification probability among the five Datasets and the appropriate classification model.
RESULTS
Among the 5 Datasets, Peak value Dataset when changing to Frequency domain through FFT showed the highest classification probability of each classification model, and NaiveBayes of each classification model had appropriate advantages for classification of pitching motion. Therefore NaiveBayes has decided on an appropriate classification model to classify pitching motion.
CONCLUSIONS
The data of the acceleration sensor and the gyroscope of the six actions are best classified for conversion using FFT and the NaiveBayes classification model is an appropriate classification model for classifying each motion. |
| Keywords:
overuse, IMU sensor, data mining, pitching, classification |
|
PDF Links
Full text via DOI
Download Citation
