DocumentCode :
3580674
Title :
Activity Estimation Using Regression Technique
Author :
Bhuvaneswari, P.T.V. ; Gayathri, S. ; Priyadharshini, A. Saraswathi
Author_Institution :
Dept. of Electron. Eng., Anna Univ., Chennai, India
fYear :
2014
Firstpage :
1177
Lastpage :
1183
Abstract :
Estimation of human activities using regression techniques has been performed in this paper. The activities considered for investigation are sitting, standing and walking. As the number of independent variable considered in the proposed work is more than two, Multiple Variate Regression (MVR) technique is applied to estimate the activities. From the performance analysis, it is found that this technique results in an overall accuracy of 99.8% of the activity sitting, 99.85% of the activity standing and 99.95% for the activity walking. However, the process involved in estimation is found to be time consuming. As the process is carried out in real time, in order to reduce the time consumption, the study is extended to Multiple Linear Regression (MLR) technique. From the analysis, it is found that MLR technique yields better result when compared to the MVR technique in terms of time consumption without compromising accuracy.
Keywords :
accelerometers; gesture recognition; regression analysis; signal classification; ubiquitous computing; MLR technique; MVR technique; activity classification; human activity estimation; multiple linear regression technique; multiple variate regression technique; sitting; standing; tri-axial accelerometer data; ubiquitous computing; walking; Accuracy; Decision support systems; Equations; Estimation; Linear regression; Mathematical model; Vectors; Accuracy and Time consumption; Activity Estimation; MLR; MVR; Regression;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computational Intelligence and Communication Networks (CICN), 2014 International Conference on
Print_ISBN :
978-1-4799-6928-9
Type :
conf
DOI :
10.1109/CICN.2014.246
Filename :
7065666
Link To Document :
بازگشت