Human activity recognition technology is growing in popularity (e.g. fitbit) and the extent to which the data collected from these devices can be used to benefit people is still being explored. This data set was collected in an effort to answer whether human activity recognition technology can be used to assess the quality with which an exercise is carried out. This data set consisted of 6 young healthy male participants (age 20-28) who wore three on-body accelerometers, one around the abdomen, another around the wrist, one around the arm, and an additional one on a dumbbell. The participants had little weight lifting experience and used a dumbbell (1.25 kg) to perform one set of 10 repetitions of the Unilateral Dumbbell Biceps Curl in five different fashions under the supervision of an experienced weight lifter.
The outcome variable of interest is the different manners in which the dumbbell curls were carried out (referred to as classe in the dataset). These include doing the curl exactly according to the specification (Class A), throwing the elbows to the front (Class B), lifting the dumbbell only halfway (Class C), lowering the dumbbell only halfway (Class D) and throwing the hips to the front (Class E). Class A is the correct manner in which the exercise should be carried out, while the other classes are common errors.
The predictors of interest include 152 predictors with information about the magnitude, acceleration, and gyroscope position of the different accelerometers in addition to summary statistics such as variance and standard deviation.
The data set can be found at the following link
Velloso, E.; Bulling, A.; Gellersen, H.; Ugulino, W.; Fuks, H. Qualitative Activity Recognition of Weight Lifting Exercises. Proceedings of 4th International Conference in Cooperation with SIGCHI (Augmented Human ’13) . Stuttgart, Germany: ACM SIGCHI, 2013.