Authors :
Safak Kayikci; Seda Postalcioglu
Volume/Issue :
Volume 5 - 2020, Issue 11 - November
Google Scholar :
http://bitly.ws/9nMw
Scribd :
https://bit.ly/3mpL65l
Abstract :
Recognition of human activities has become a
very popular problem that has been widely studied with
the development of sensors embedded in mobile devices
and increasingly widespread methods of machine
learning. For the solution of this problem, the sensor
data of the different movements collected are labeled
with the movements performed and turned into a
classification problem. Different human activities are
tried to be distinguished by applying Gradient Boosting,
Random Forest, AdaBoost, Gaussian Naive Bayes
classification algorithms on the collected data.
Performance examinations and accuracy values are
evaluated with the combination confusion matrix. It is
observed that Gradient Boosting showed the best
performance overall analysis. Human activity
recognition is used in health practices, calculating
personal daily calories, analyzing the health status,
monitoring the movements performed by the elderly
people in their environment, human position tracking,
and various security applications
Keywords :
Human activity recognition; Sensor; Gradient Boosting; Random Forest; AdaBoost; Gaussian Naive Bayes.
Recognition of human activities has become a
very popular problem that has been widely studied with
the development of sensors embedded in mobile devices
and increasingly widespread methods of machine
learning. For the solution of this problem, the sensor
data of the different movements collected are labeled
with the movements performed and turned into a
classification problem. Different human activities are
tried to be distinguished by applying Gradient Boosting,
Random Forest, AdaBoost, Gaussian Naive Bayes
classification algorithms on the collected data.
Performance examinations and accuracy values are
evaluated with the combination confusion matrix. It is
observed that Gradient Boosting showed the best
performance overall analysis. Human activity
recognition is used in health practices, calculating
personal daily calories, analyzing the health status,
monitoring the movements performed by the elderly
people in their environment, human position tracking,
and various security applications
Keywords :
Human activity recognition; Sensor; Gradient Boosting; Random Forest; AdaBoost; Gaussian Naive Bayes.
