The Human Activity Recognition database was built from the recordings of 15 study participants performing activities of daily living (ADL) while carrying a waist-mounted smartphone with embedded inertial sensors. The objective is to classify activities into one of the four activities performed.
The last columns of the train.csv file, a class label The rest of the columns contain the sensor values of the associated activity. The label can be:
LAYING:0
SITTING:1
STANDING:2
WALKING:3
The sensor signals (accelerometer and gyroscope) were pre-processed by applying noise filters and then sampled in fixed-width sliding windows of 2.56 sec and 50% overlap (128 readings/window). The sensor acceleration signal, which has gravitational and body motion components, was separated using a Butterworth low-pass filter into body acceleration and gravity. The gravitational force is assumed to have only low-frequency components; therefore, a filter with a 0.3 Hz cutoff frequency was used. From each window, a vector of features was obtained by calculating variables from the time and frequency domain.
The evaluation metric for this contest is the categorization accuracy or the proportion of activities that are correctly classified. For example, a categorization accuracy of 0.97 indicates that you have correctly classified all but 3% of the activity.