Physical Activity Monitoring

You are employed as a data scientist at Trivisio. At the moment we are trying to sell the new Colibri Wireless unit, which is an inertial measurement unit $($IMU$).$ It is made up of $3$ sensors to measure acceleration, angular rate and magnetic field, and separates sensor to measure temperature and orientation. We want our product to be superior to our main competitors: Wireless InertiaCube$3$ by InterSense, MTw Awinda by Xsens, T$40$s by Vicon, and OptoTrak $3020$ by Optotrak Certus.

We have collected data from $9$ subjects who wore $3$ IMUs and a heart$-$rate monitor. They each conducted $12$ different physical activities $($e$.$g$.,$ walking, cycling, playing football$),$ with some conducting an additional $6$ extra activities. The data is stored in individual $.$dat files per each subject. Each row in the dataset represents a single reading on the companion unit $($a dongle that plugs into a computer to act as a receiver for the data$),$ containing $54$ attributes:

timestamp

activityID

heart$-$rate bpm $($beats per minute$)$

$(4-20)$ IMU readings from the unit attached to the wrist on the dominant arm

$(21-37)$ IMU readings from the unit attached to the chest

$(38-54)$ IMU readings from the unit attached to the ankle on the dominant leg