Human Activity Recognition (HAR) has wide applications in healthcare (remote monitoring of patients), smart environments (smart homes, IoT etc.), sports (fitness applications and monitoring), and many more.
Nowadays, where nearly every smartphone (or smart watch) comes equipped with built in inertial sensors, such as accelerometers and gyroscopes, the technology for HAR is available for almost everybody.
Our goal is to built an HAR classification system, using accelerometer and gyroscope data generated by the user's cell phone.
The data is taken from the Human Activity Recognition database, built from the recordings of 30 subjects performing activities1 of daily living while carrying a waist-mounted smartphone with embedded inertial sensors.2
We will try and test different predictive algorithms, and estimate the models' accuracy on an independent test set.
1: Follow this link or simply click on the above image in order to watch a video features one of the participants performing the 6 activities recorded in the experiment.
2: The data is available at the UC Irvine Machine Learning Repository (link to data)